Beruflich Dokumente
Kultur Dokumente
Editors:
J.-M. Morel, Cachan
F. Takens, Groningen
B. Teissier, Paris
Jean-Paul Brasselet · José Seade · Tatsuo Suwa
Vector Fields
on Singular Varieties
123
Jean-Paul Brasselet Tatsuo Suwa
IML-CNRS, Case 907, Luminy Department of Mathematics
13288 Marseille Cedex 9, France Hokkaido University
jpb@iml.univ-mrs.fr Sapporo 060-0810, Japan
tsuwa@sci.hokudai.ac.jp
José Seade
Instituto de Matemáticas
Unidad Cuernavaca
Universidad Nacional Autónoma de México
C.P. 62210, Cuernavaca, Morelos, México
jseade@matcuer.unam.mx
Mathematics Subject Classification (2000): 32S65, 37F75, 57R20, 57R25, 58K45, 14B05, 14C17, 14J17,
32S05, 32S55, 58K65
The use of general descriptive names, registered names, trademarks, etc. in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant protective
laws and regulations and therefore free for general use.
springer.com
Preface
Vector fields on manifolds play major roles in mathematics and other sciences.
In particular, the Poincaré–Hopf index theorem and its geometric counter-
part, the Gauss–Bonnet theorem, give rise to the theory of Chern classes, key
invariants of manifolds in geometry and topology.
One has often to face problems where the underlying space is no more a
manifold but a singular variety. Thus it is natural to ask what is the “good”
notion of index of a vector field, and of Chern classes, if the space acquires sin-
gularities. The question was explored by several authors with various answers,
starting with the pioneering work of M.-H. Schwartz and R. MacPherson.
We present these notions in the framework of the obstruction theory and
the Chern–Weil theory. The interplay between these two methods is one of
the main features of the monograph.
v
Acknowledgements
Parts of this monograph were written while the authors were staying at
various institutions, such as Hokkaido University and Niigata University in
Japan, CIRM, Université de la Mediterranée and IML at Marseille, France,
the Instituto de Matemáticas of UNAM at Cuernavaca, Mexico, ICTP at
Trieste, Italia, IMPA at Rio de Janeiro, and USP at São Carlos in Brasil, to
name a few, and we would like to thank them for their generous hospitality
and support.
Thanks are also due to people who helped us in many ways, in particular
our co-authors of results quoted in the book: Marcelo Aguilar, Wolfgang
Ebeling, Xavier Gómez-Mont, Sabir Gusein-Zade, Lê Dũng Tráng, Daniel
Lehmann, David Massey, A.J. Parameswaran, Marcio Soares, Mihai Tibar,
Alberto Verjovsky, and many other colleagues who helped us in various ways.
We are also grateful to CNRS (France), CONACYT and UNAM-DGAPA
(Mexico) and JSPS (Japan) for financial support.
Thanks to Vincent Blanloeil, Institut de Recherche Mathématique
Avancée, Strasbourg, France, who allowed us to use the picture of Barth
sextic in cover.
We would like to express our indebtedness to M.-H. Schwartz for her great
contribution to the theory, that has played and keeps playing an important
role in our mathematics.
vii
Contents
ix
x Contents
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
Introduction
The study of vector fields and flows near an isolated singularity, or stationary
point, has played for decades, and even centuries, a major role in several areas
of mathematics and in other sciences, notably in physics, biology, economics,
etc. The most basic invariant of a vector field at an isolated singularity is its
local Poincaré–Hopf index, which has been studied from very many different
viewpoints and there is a vast literature about it. At the same time, it is be-
coming more and more usual to face problems and situations in Mathematics
(and in other sciences) where the underlying space is not a manifold but a
singular variety. It is thus natural to ask what should be the “good” notion
of index of a vector field on a singular variety, depending on which properties
of the local index we have in mind.
For instance one has the theorem of Poincaré–Hopf saying that the sum
of the local indices of a vector field with isolated singularities on a closed
oriented manifold, is independent of the choice of the vector field and equals
the Euler–Poincaré characteristic of the manifold in question. Defining an
index for vector fields on singular varieties that has this property leads to the
Schwartz index, that we explain below.
Similarly, an important property of the local Poincaré–Hopf index is that
it is stable under perturbations, or in other words, that if we approximate the
given vector field by another vector field which has only Morse singularities,
then the local index of the initial vector field is the number of singularities
of its morsification counted with signs. Defining an index for vector fields on
singular varieties that has this property leads to a different index, the GSV
index.
There are other important properties of the local Poincaré–Hopf index
that give rise to various other indices when we look at singular varieties.
That makes the study of indices of vector fields over singular varieties an
interesting field of current research, which combines an amazing variety of
ideas and techniques coming from algebraic topology, differential geometry,
algebraic geometry, dynamical systems, mathematical physics, etc.
The goal of this monograph is to give an account of the various indices
of vector fields on singular varieties that are in the literature, the relations
among them, and the way how they relate with various generalizations of
xiii
xiv Introduction
Chern classes to singular varieties. Indices of vector fields and Chern classes
of vector bundles are nowadays present in many branches of mathematics,
and these two concepts are linked together in an essential way.
This monograph goes together with [28] to give a global view of the theory
of indices and Chern classes for singular spaces. In [28] the focus is on the
theory of characteristic classes for singular varieties. Here the emphasis is on
indices and their relation with Chern classes. We do this following two of the
classical viewpoints for studying Chern classes, both introduced by Chern
himself. These are the topological viewpoint, thinking of Chern classes as
being the primary obstruction to constructing sections of appropriate fiber
bundles, and the differential-geometric viewpoint, via Chern–Weil theory,
where the corresponding classes are localized at the “singularities” of certain
connections via the theory of residues, which is largely indebted to R. Bott.
The interplay between these two viewpoints for studying indices and char-
acteristic classes, obstruction theory and Chern–Weil theory, is a key feature
of this monograph.
This work does not pretend to be comprehensive, and yet it offers a global
viewpoint of the theory of indices of vector fields and Chern classes of singular
varieties that can be of interest for people working in singularities, algebraic
and differential geometry, algebraic topology, and even in string theory and
mathematical physics. In each individual chapter we indicate additional ref-
erences to the literature, for further reading.
The study of indices of vector fields and Chern classes for singular vari-
eties started in the early 1960s with M.-H. Schwartz, and then continued by
R. MacPherson and many others. This is today an active field of research,
in which the foundations of the theory are being laid out by several authors,
and so are their relations with other branches of geometry, topology, and
singularity theory.
We start Chap. 1 with the basic, well-known, theory of indices of vector
fields and Chern classes that we need in the sequel, and we describe for
manifolds the two viewpoints that we use in the rest of the work to study
these invariants, namely localization via obstruction theory and localization
via Chern–Weil theory.
In Chap. 2 begins the discussion of indices of vector fields on singular
varieties. We start with the index introduced by M.-H. Schwartz (in [139,141])
in her study of Chern classes for singular varieties. For her purpose there was
no point in considering vector fields in general, but only a special class of
vector fields (and frames) that she called “radial,” which are obtained by the
important process of radial extension that she introduced. The generalization
of this index to other vector fields was first done by H. King and D. Trotman
in [96], and later independently in [6,49,149]. We call this the Schwartz index;
in the literature it is also called “radial index” because it measures how far
the vector field is from being radial. The corresponding discussion for frames
is done in Chap. 10.
Introduction xv
0 −→ ΩV,0
n
−→ ΩV,0
n−1
−→ · · · −→ OV,0 −→ 0 ,
where the arrows are given by contracting forms by the vector field v. The
homological index of v is defined to be the Euler characteristic of this com-
plex. If the ambient space V is smooth at 0, the complex is exact in all
dimensions, except in degree 0 where the corresponding homology group has
dimension equal to the local Poincaré–Hopf index of v at 0. If (V, 0) is an
ICIS, the recent article [17] of H.-C. Graf von Bothmer, W. Ebeling and X.
Gómez-Mont shows that this index coincides with the GSV index, a fact
previously known only for vector fields on hypersurface germs. We remark
however that the homological index is defined for vector fields on arbitrary
isolated normal singularity germs, while the GSV index is only defined on
complete intersection germs. Hence the homological index does provide a
new invariant for singular varieties which is not yet understood in general. It
would be interesting to know what this index measures globally, i.e., given a
compact variety W with isolated singularities and a holomorphic vector field
on it with isolated singularities, its total homological index an invariant of
W . What type of invariant is it? If W is a local complete intersection, this
is just the usual Euler–Poincaré characteristic of a smoothing of W , and as
explained in the text, this equals the 0-degree Fulton–Johnson class of W .
Introduction xvii
part Vreg , isomorphic to the normal bundle. One has an isomorphism (as C ∞
vector bundles)
T M |Vreg = T Vreg ⊕ N |Vreg ,
and therefore the virtual bundle τV = [T M |V − N |V ], regarded as an element
in the K-theory group KU (V ), is called the virtual tangent bundle of V .
The homology Chern classes of the virtual tangent bundle τV are the Fulton–
Johnson classes of V . In this book we envisage only the case, where V is a
local complete intersection in the complex manifold M . When localized at
the singular set of a vector field, the local contribution to the 0-dimensional
part of the Fulton–Johnson class is the virtual index. When V has only
isolated singularities, this corresponds to the Euler–Poincaré characteristic
of a smoothing of V . The generalization of the virtual index for frames and
its relation with Chern classes are given in Chap. 11.
In general, these classes are different from the Schwartz–MacPherson
classes. If V has only isolated singularities, then (by [149,155]) the Schwartz–
MacPherson and Fulton–Johnson classes coincide in all dimensions other than
0, and in dimension 0 this difference is given by the local Milnor numbers of V
at its singular points. Hence it is natural to call Milnor classes the difference
between Fulton–Johnson and Schwartz–MacPherson classes. These classes
were studied by P. Aluffi [8], who called them μ-classes; there have been sig-
nificant contributions to the subject afterwards, either by Aluffi himself and
by various other authors, such as S. Yokura, A. Parusiński and P. Pragacz,
D. Lehmann, T. Ohmoto, J. Schürmann, and the authors of this monograph.
This is studied in Chap. 12.
Of course one may also compare Chern–Mather with Fulton–Johnson
classes. This was done in [125] for (strong) local complete intersections with
isolated singularities, using results of [149,155]. As in the previous case, these
classes coincide in all dimensions greater than 0; in dimension 0 their differ-
ence is given by the polar multiplicities of T. Gaffney. The corresponding
study for varieties with nonisolated singularities has not been done yet.
Finally, the fourth way for extending the concept of tangent bundle to sin-
gular varieties by considering the tangent sheaf ΘV , which is by definition the
dual of ΩV , the sheaf of Kähler differentials on V , fits within the framework
considered in [158] of Chern classes for coherent sheaves. We briefly describe
some of their properties in Chap. 13. In particular, if V is a local complete
intersection in M , then one has a canonical locally free resolution of ΩV and
the corresponding Chern classes essentially coincide with the Fulton–Johnson
classes, though the corresponding classes for ΘV differ from these.
In the sequel we explain how the various indices of vector fields that we
discuss in Chaps. 2–8 are related among themselves and how they relate to
some generalization of the Chern classes of manifolds to the case of singular
varieties. There is however something missing in this picture: so far we do
not know of a direct relation between the homological index and some type
of Chern classes for singular varieties, neither we know of a direct relation
between the Chern classes of the tangent sheaf (or its dual) and some index
of vector fields (or 1-forms).
xx Introduction
While writing this monograph we have tried to convey the reader a unified
view of the various generalizations for singular varieties one has of the im-
portant concepts of the local index of Poincaré–Hopf and Chern classes of
manifolds. These are topics of current research which keep developing and
the literature is vast, so we focused on the most classical approaches for this
subject. There are of course important topics that were just glanced here,
or maybe even not discussed at all, specially concerning new trends in alge-
braic geometry and topology, such as string theory and motivic integration.
Yet, we think the content of this monograph contributes to lay down the
foundations of a theory for singular varieties which is just beginning to be
developed and understood. This ought to play in the future such an impor-
tant role for understanding the geometry and topology of singular varieties
as they do for manifolds. And this should also have important applications to
other branches of knowledge, where it is important to consider vector fields
and flows on orbifolds and singular varieties.
Chapter 1
The Case of Manifolds
IndPH (v, M ),
In the following, singularities of the vector field v are not necessarily isolated
points. We still define a Poincaré–Hopf index in that case.
4 1 The Case of Manifolds
This number IndPH (v, S) depends only on the behavior of v near S and
not on the choice of the neighborhood U , or of the tube T . Moreover, for
this index it does not matter what actually happens on S, we only care what
happens around S, but away from S. In particular, if v is “radial” from S,
i.e., if it is transverse to the boundary of a cellular tube around S pointing
outward, then IndPH (v, S) = χ(S).
Now let M be a compact oriented C ∞ manifold possibly with boundary
∂M and v a continuous vector field on M , nonsingular on the boundary. From
1.2 Poincaré and Alexander Dualities 5
the above considerations, we may assume that the set S(v) of singular points
of v has only a finite number of components {Sλ }.
If M has no boundary, the Poincaré–Hopf Theorem implies that
IndPH (v, Sλ ) = χ(M ). (1.1.3)
λ
If M has a boundary, the sum λ IndPH (v, Sλ ) depends only on the
behavior of v near ∂M . For example, if v is pointing outwards everywhere
on ∂M , then we have the same formula (1.1.3). If v is pointing inwards
everywhere on ∂M , the right hand side becomes χ(M )−χ(∂M ). In particular,
if the (real) dimension of M is even (as it will usually be the case in this book)
and if v is everywhere transverse to ∂M , then we have again the same formula
(1.1.3).
Here we introduce the concept of the difference which will be used in
the rest of the book. For this we let v and v be continuous vector fields
on a neighborhood U of S in M , nonsingular on U \ S. Let T and T be
cellular tubes around S in U such that interior of T contains the closure of T
and denote X = T \ T . Let us consider w a vector field on X with isolated
singularities which restricts to v on ∂T and to v on ∂T ; such a vector field
w always exists by Theorem 1.1.2. We may denote by d(v, v ) = IndPH (w, X)
the difference between v and v . Then one has:
One can easily prove the following result that will be used later.
We briefly review the classical case, which will be generalized to the case of
singular varieties in Sect. 10.4 below. In either case, we follow the descriptions
given in [25].
Let M be an oriented manifold of real dimension m . We take a triangu-
lation (K) of M and the cellular decomposition (D) dual to (K), as before.
(K)
The groups of chains relative to (K) and (D) are denoted by C∗ (M ) and
(D)
C∗ (M ), respectively. Also, the groups of cochains relative to (K) and (D)
∗ ∗
are denoted by C(K) (M ) and C(D) (M ), respectively. The intersection of an
6 1 The Case of Manifolds
i-simplex σ and its dual (m − i)-cell d(σ) is transverse and consists of one
point, the barycenter σ̂ of σ.
First, if M is compact, we define a homomorphism
m −i (K)
P : C(D) (M ) −→ Ci (M ) by P (c) = c, d(σ) σ (1.2.1)
σ
for an (m − i)-cochain c, where the sum is taken over all i-simplices σ of
M (we follow the orientation conventions in [25]). This induces the Poincaré
isomorphism ∼
PM : H m −i (M ) −→ Hi (M ).
Next, let S be a (K)-subcomplex of M whose geometric realization is also
∗ ∗
denoted by S. Let C(D) (M, M \S) denote the subgroup of C(D) (M ) consisting
of cochains which are zero on the cells not intersecting with S.
Suppose S is compact (M may not be compact). Then we may define a
homomorphism
m −i (K)
A : C(D) (M, M \ S) −→ Ci (S)
taking, in the sum in (1.2.1), only i-simplices of S. This induces the Alexander
isomorphism ∼
AM,S : H m −i (M, M \ S) −→ Hi (S).
From the construction, we have the following
j∗
H m −i (M, M \ S) −−−−→ H m −i (M )
⏐ ⏐
⏐
AM,S
⏐
PM
i
Hi (S) −−−∗−→ Hi (M ).
Let us recall the definition of the Chern classes via obstruction theory [28,
89, 123, 153]. This can be done in full generality, however for simplicity we
consider first the case of Chern classes of almost-complex manifolds, and later
in this section we indicate how this generalizes to complex vector bundles in
general.
Now we assume we are given an almost complex m = 2m-manifold M ,
so its tangent bundle T M is endowed with the structure of a complex vector
bundle of rank m.
1.3 Chern Classes via Obstruction Theory 7
v (r) pr2
∂σ
Sk−1 −→ Wr (T M )|U
U × Wr,m −→ Wr,m ,
There is another useful definition of the index Ind(v (r) , σ): let us write
the frame v (r) as (v (r−1) , vr ), where the last vector is individualized, and
suppose that v (r) is already defined on ∂σ. There is no obstruction to ex-
tending the (r − 1)-frame v (r−1) from ∂σ to σ because the dimension of the
obstruction for such an extension is 2(m − (r − 1) + 1) = dim σ + 2. The
(r − 1)-frame v (r−1) , defined on σ, generates a complex subbundle Gr−1 of
rank (r − 1) of T M |σ and one can write
T M |σ
Gr−1 ⊕ Qq ,
Remark 1.3.1. The Chern classes of complex vector bundles in general are
defined in essentially the same way as above. If E is a complex vector bundle
of rank k > 0 over a locally finite simplicial complex B of dimension n ≥ k,
then one has Chern classes ci (E) ∈ H 2i (B; Z), i = 1, . . . , k. The class ci (E)
is by definition the primary obstruction to constructing (k − i + 1) linearly
independent sections of E.
The class c0 (E) is defined to be 1 and one has the total Chern class of E
defined by:
c∗ (E) = 1 + c1 (E) + · · · + ck (E)
This can be regarded as an element in the cohomology ring H ∗ (B) and it is
invertible in this ring.
For r = 1 the frames consist of a single vector field and the difference above
corresponds, via Poincaré duality, to the one previously defined for vector
fields (cf. 1.1.4).
In the sequel, we will show that the relative Chern class allows us to define
Chern class in homology.
Let S be a compact (K)-subcomplex of M , and U a neighborhood of S.
Let T be a cellular tube in U around S. Take an r-field v (r) defined on D(2q) ,
possibly with singularities. We suppose that the only singularities inside U
are located in S. This implies that v (r) has no singularities on (∂T )(2q) so
there is a well defined relative Chern class (see 1.3.3)
cq (T , ∂T ; v (r) ) ∈ H 2q (T , ∂T ).
10 1 The Case of Manifolds
For r = 1 the frame consists of a single vector field v and the class
PH(v, S) ∈ H0 (S) is identified with the Poincaré–Hopf index of v at S,
IndPH (v, S), previously defined (Definition 1.1.3).
Note that if dim S < 2r − 2, then PH(v (r) , S) = 0.
The relation between the Poincaré–Hopf class of v (r) and the index we
defined above is the following:
PH(v (r) , S) = Ind(v (r) , d(σ)) σ ,
where the sum runs over the 2(r − 1)-simplices σ of the triangulation of S
and d(σ) is the dual cell of σ (of dimension 2q).
Let us consider now the case of manifolds with boundary. Let M be a
compact almost complex 2m-manifold, with nonempty boundary ∂M . Let
(K) be a triangulation of M compatible with ∂M . The union of all “half-cells”
dual to simplices in ∂M , denoted by U is a regular neighborhood of ∂M . Its
boundary is denoted by ∂U, which is a union of (D)-cells and is homeomorphic
to ∂M . The pair (M \ (Int U), ∂U) is homeomorphic to (M, ∂M ) and one can
apply the previous construction.
Let v (r) be an r-field on the (2q)-skeleton of (D), with singularities located
on a compact subcomplex S in M \ (Int U). On the (2q)-skeleton of U, we
have a well defined r-frame v (r) . Let {Sλ } be the connected components of
S. Then, by setting cr−1 (M ; v (r) ) = cq (M, ∂M ; v (r) ) [M, ∂M ], we have
(iλ )∗ PH(v (r) , Sλ ) = cr−1 (M ; v (r) ) in H2r−2 (M ), (1.3.3)
λ
Remark 1.3.2. In the sequel we often speak of localizing Chern classes, which
can be done by two different methods: either obstruction theory or Chern–
Weil theory. The obstruction theoretical viewpoint comes from the above
concept of relative Chern classes: if S is a compact sub-complex of M , U a
1.4 Chern–Weil Theory of Characteristic Classes 11
∇ : A0 (M, E) −→ A1 (M, E)
d : A0 (M ) −→ A1 (M )
∇1 , . . . , ∇ be connections
Lemma 1.4.2. Let for E and f1 , . . . , f C ∞ func-
tions on M with i=1 fi ≡ 1. Then i=1 fi ∇i is a connection for E.
One of the consequences of the above lemmas is that every vector bundle
admits a connection. This can be shown by taking an open covering U of M
so that E is trivial on each open set in U, choosing a connection on each open
set trivial with respect to some frame of E, and then patching them together
by a partition of unity subordinate to U.
If ∇ is a connection for E, it induces a C-linear map
∇ : A1 (M, E) −→ A2 (M, E)
satisfying
The composition
K = ∇ ◦ ∇ : A0 (M, E) −→ A2 (M, E)
The fact that a connection is a local operator allows us to obtain local rep-
resentations of it and its curvature by matrices whose entries are differential
forms. Thus suppose that ∇ is a connection for a vector bundle E of rank k
and that E is trivial on U . If e = (e1 , . . . , ek ) is a frame of E on U , we may
write, for i = 1, . . . , k,
k
∇(ei ) = θji ⊗ ej , θji ∈ A1 (U ).
j=1
We call θ = (θij ), the matrix whose (i, j) entry is θij , the connection matrix
k∇ with respect to e. For an
of
∞
arbitrary section s on U , we may write s =
i=1 fi ei where the fi are C functions on U and we compute
1.4 Chern–Weil Theory of Characteristic Classes 13
k
k
∇(s) = (dfi + θij fj ) ⊗ ei .
i=1 j=1
k
k
K(ei ) = κji ⊗ ej , κij = dθij + θi ∧ θj .
j=1 =1
Let m = [m /2] and, for each i = 1, . . . , m, let σi denote the ith elemen-
tary symmetric function in m variables X1 , . . . , Xm , i.e., σi (X1 , . . . , Xm ) is
a polynomial of degree i defined by
m
(1 + Xi ) = 1 + σ1 (X1 , . . . , Xm ) + · · · + σm (X1 , . . . , Xm ).
i=1
Since differential forms of even degrees commute with one another with re-
spect to the exterior product, we may treat κ as an ordinary matrix whose
entries are numbers. We define a 2i-form σi (κ) on U by
where I denotes the identity matrix of rank k. Note that σi (κ) = 0 for
i = k + 1, . . . , m, and in particular, σ1 (κ) is the trace tr(κ) and σk (κ) is the
determinant det(κ). Although σi (κ) depends on the connection ∇, by (1.4.1),
it does not depend on the choice of the frame of E and it defines a global
2i-form on M , which we denote by σi (∇). It is shown that the form is closed
([75, Ch.3, 3 Lemma], [123, Appendix C, Fundamental Lemma]). We set
√ i
i −1
c (∇) = σi (∇)
2π
= (1 − t)∇ + t∇ ,
∇
Then we set
ci (∇, ∇ ) = π∗ (ci (∇)). (1.4.3)
A similar construction works for an arbitrary collection of finite number of
connections and the resulting differential form is called the Bott difference
form ([19, p. 65]).
From the above, we see that the class [ci (∇)] of the closed 2i-form ci (∇)
in the de Rham cohomology H 2i (M, C) depends only on E and not on the
choice of the connection ∇. We denote this class by ci (E) and call it the i-th
Chern class ci (E) of E via the Chern–Weil theory. We call
Remark 1.4.1. 1. It is known (see, e.g., [123]) that the class ci (E) defined as
above is the image of the class ci (E) in H 2i (M, Z) defined via the obstruction
theory by the canonical homomorphism
This fact can also be proved directly using an expression of the mapping
degree in terms of connections (see, e.g., [161]).
2. Let H be a hyperplane in the projective space CPm . For the hyperplane
bundle LH , the line bundle determined by H, we have
c(LH ) = 1 + hm ,
D(p−1) D(p)
· · · −→ Ap−1 (U) −→ Ap (U) −→ Ap+1 (U) −→ · · ·
Now we recall the integration on the Čech-de Rham cohomology (cf. [109]).
Suppose that the m -dimensional manifold M is oriented and compact and
let U = {U0 , U1 } be a covering of M . Let R0 , R1 ⊂ M be two compact
manifolds of dimension m with C ∞ boundary with the following properties:
(1) Rj ⊂ Uj for j = 0, 1,
(2) IntR0 ∩ IntR1 = ∅ and
(3) R0 ∪ R1 = M .
Let R01 = R0 ∩ R1 and give R01 the orientation as the boundary of R0 ;
R01 = ∂R0 , equivalently give R01 the orientation opposite to that of the
boundary of R1 ; R01 = −∂R1 . We define the integration
: Am (U) −→ C by ξ= ξ0 + ξ1 + ξ01 .
M M R0 R1 R01
Then by the Stokes theorem, we see that if Dξ = 0 then M ξ is
independent of {R0 , R1 } and that if ξ = Dη for some η ∈ A (U) then
p−1
M
ξ = 0. Thus we may define the integration
m
: HD (U) −→ C,
M
1.5 Čech-de Rham Cohomology 17
which is compatible with the integration on the de Rham cohomology via the
isomorphism of 1.5.1.
Next we define the relative Čech-de Rham cohomology and describe the
Alexander duality. Let M be an m -dimensional oriented manifold (not nec-
essarily compact) and S a compact subset of M . Let U0 = M \ S and let U1
be an open neighborhood of S. We consider the covering U = {U0 , U1 } of M .
We set
Ap (U, U0 ) = { ξ = (ξ0 , ξ1 , ξ01 ) ∈ Ap (U) | ξ0 = 0 }.
Then we see that if ξ is in Ap (U, U0 ), Dξ is in Ap+1 (U, U0 ). This gives rise
to another complex, called the relative Čech-de Rham complex, and we may
define the p-th relative Čech-de Rham cohomology of the pair (U, U0 ) as
p
HD (U, U0 ) = KerDp /ImDp−1 .
m
and induces an operator M : HD (U, U0 ) → C.
In the cup product A (U) × Am −p (U) → Am (U) given as (1.5.2), we
p
see that if ξ0 = 0, the right hand side depends only on ξ1 , ξ01 , and η1 . Thus
we have a pairing Ap (U, U0 ) × Am −p (U1 ) → Am (U, U0 ), which, followed by
the integration, gives a bilinear pairing
Ap (U, U0 ) × Am −p (U1 ) −→ C.
j∗
H p (M, M \ S; C) −−−−→ H p (M, C)
⏐ ⏐
⏐
A
⏐
P
i
Hm −p (S, C) −−−∗−→ Hm −p (M, C),
p−1 p
HdR (U0 ) −→ HD (U, U0 ) given by [ω] → [(0, 0, −ω)]
can be shown to be an isomorphism (we leave the details to the reader) and
we have
p p−1 p−1 m −1 C, for p = m ,
HD (U, U0 )
HdR (U0 )
H (S )=
0, for p = 2, . . . , m − 1.
An explicit generator of H m −1 (Sm −1 ) is given as follows ([75, p. 370]).
For x = (x1 , . . . , xm ) in Rm , we set Φ(x) = dx1 ∧ · · · ∧ dxm and
i ∧ · · · ∧ dxm .
Φi (x) = (−1)i−1 xi dx1 ∧ · · · ∧ dx
1 dz
β1 = √ ,
2π −1 z
Then we see that Dci (∇∗ ) = 0 and this defines a class [ci (∇∗ )] in HD
2i
(U). It
is not difficult to show the following
Theorem 1.6.2. The class [ci (∇∗ )] ∈ HD
2i
(U) corresponds to the Chern class
ci (E) ∈ HdR
2i
(M ) under the isomorphism of Theorem 1.5.1.
By a similar construction, we may define the characteristic class ϕ(E) for
a polynomial ϕ in the Chern polynomials in the Čech-de Rham cohomology.
It can be done also for virtual bundles (see Chap. 5).
Using Bott difference forms, we may define characteristic classes in the
Čech-de Rham cohomology for an arbitrary open covering of M .
This way of representing characteristic classes is particularly useful in deal-
ing with the “localization problem,” which we explain in the next subsection.
This theory involves vanishing theorems, one of which is given as follows.
Let E be a complex vector bundle of rank k on a C ∞ manifold M . Let
s = (s1 , . . . , sr ) be an r-frame of E on an open set U . Recall that (Definition
1.4.2) a connection ∇ for E on U is s-trivial, if ∇(si ) = 0 for i = 1, . . . , r.
Proposition 1.6.1. If ∇ is s-trivial, then
cj (∇) ≡ 0 for j ≥ k − r + 1.
ϕ(∇0 ) ≡ 0.
and accordingly, we have the residue Resϕ (γ, E; Sλ ) in Hm −2d (Sλ , C) for
each λ. Replacing U1 by a regular neighborhood Uλ of Sλ , disjoint from the
other components, and R1 by an m -dimensional manifold Rλ with boundary
in Uλ containing Sλ in its interior, we have an expression (1.6.3)λ or (1.6.4)λ
for the residue Resϕ (γ, E; Sλ ) similar to (1.6.3) or (1.6.4).
From the above considerations and Proposition 1.5.1, we have the following
“residue theorem.”
ck (∇) = 0. (1.6.6)
Remark 1.6.2. 1. In fact it can be shown that the above residues are in the
integral homology and the equality in Theorem 1.6.7 holds in the integral
homology (see [161]).
2. A localization theory of Chern classes, other than the top one, by a finite
number of sections can be developed similarly (see [159–161]).
where (ρ01 )∗ denotes the integration along the fiber of ρ01 so that the form
(ρ01 )∗ c1 (∇0 , ∇1 ) is in fact a function on H. It suffices to prove that this
function is identically equal to −1. Let p be an arbitrary point in H and
suppose it is in Wi , i = 0. In the sequel, we identify LH |Wi with Wi × C. On
Wi , the section s is represented by the function si = ζ0 /ζi , which can also
be thought of as a fiber coordinate of the retraction ρ. Let ∇ denote the
connection for LH |H on Wi ∩ H trivial with respect to the frame given by
(q) = (q, 1) for q in Wi ∩ H. We may modify ∇ away from a neighborhood
of p to obtain a connection ∇ for LH |H on H. The pullback ∇1 = ρ∗ ∇
is a connection for LH which is trivial with respect to the frame 1 given
by 1 (q) = (q, 1) for q in a neighborhood W of p in Wi . Now we try to find
c1 (∇0 , ∇1 ) on W ∩U01 = W \H (cf. (1.4.3)). For this, let ∇˜ be the connection
for LH × R given by ∇ = (1 − t)∇0 + t∇1 . Let θi be the connection form of ∇i
˜
with respect to the frame 1 , i = 0, 1. Then θ1 = 0 and, since θ0 is s-trivial
and 1 = 1z s, z = ζ0 /ζi , by (1.4.1), we have θ0 = d z1 / z1 = − dz
z . Thus the
connection form θ̃ of ∇ ˜ is given by
dz
θ̃ = −(1 − t) .
z
1.6 Localization of Chern Classes 25
where π∗ denotes the integration along the fiber of the projection map π :
W \ H × [0, 1] → W \ H. Therefore, by the Cauchy integral formula, we have
(ρ01 )∗ c1 (∇0 , ∇1 ) = −1
in a neighborhood of p.
See [157] and [161] for more general results and thorough discussions in
this direction.
As we have seen in the previous subsection and will see also in the sequel, the
residues of characteristic classes are deeply related to Grothendieck residues.
In this subsection, we briefly review this subject. For details, we refer to, e.g.,
[75].
Let U be a neighborhood of the origin 0 in Cm and f1 , . . . , fm holomorphic
functions on U such that their common set of zeros consists only of 0. For a
holomorphic m-form ω on U , we set
ω 1 ω
Res0 = √ , (1.6.10)
f 1 , . . . , fm (2π −1)m Γ f1 · · · fm
for a small positive number ε. We orient Γ so that the form dθ1 ∧ · · · ∧ dθm
is positive, θi = argfi .
Example 1.6.1. If m = 1, the above residue 1.6.10 is the usual Cauchy residue
at 0 of the meromorphic 1-form ω/f1 .
Example 1.6.2. In the next subsection, we give various expressions for the
residue of the top Chern class at an isolated singularity of a section s.
If (f1 , . . . , fm ) denote local components of s around the singularity, the
Grothendieck residue with ω = df1 ∧ · · · ∧ dfm appears as an “analytic ex-
pression” of the residue. Thus we have
26 1 The Case of Manifolds
df1 ∧ · · · ∧ dfm
Res0 = dimC Om /(f1 , . . . , fm ) = IndPH (v, 0), (1.6.11)
f 1 , . . . , fm
m
where v denotes the holomorphic vector field i=1 fi · ∂/∂zi . This positive
integer is also interpreted as the intersection number (D1 · · · Dm )0 at 0 of the
divisors Di defined by fi (cf. [75, Ch.5, 2], [157]).
Example 1.6.3. In particular, if fi = ∂f /∂zi for some f in Om , then the
residue is the Milnor number μ(V, 0) of the hypersurface V defined by f at 0;
∂f
d ∂z ∧ · · · ∧ d ∂f
∂zm
Res0 1
∂f ∂f
= μ(V, 0).
∂z1 , . . . , ∂zm
Proof. We indicate the proof for the case m = 1 (for m > 1, we use the
Čech-de Rham cohomology theory for m open sets, see [157], [160]). Thus
s = f e1 for some holomorphic function f on U . Let R be a closed disk about
p in U . In the expression (1.6.4) of the residue, we may take as ∇1 an e1 -trivial
connection on U , thus c1 (∇1 ) ≡ 0 and
Resp (s, E; p) = − c1 (∇0 , ∇1 )
∂R
1 df
c1 (∇0 , ∇1 ) = − √ ,
2π −1 f
Remark 1.6.3. For general m, if we take suitable connections we see that the
difference form is given by
cm (∇0 , ∇1 ) = −f ∗ βm ,
This can be proved, for example, by perturbing the sections and using the
theory of Cohen–Macaulay rings (e.g., [160]).
This can also be proved by perturbing the sections, see [75], [160].
Remark 1.6.4. There are similar expressions as above for the residues of
vector bundles on singular varieties with respect to an appropriate number
of sections (see [160]).
28 1 The Case of Manifolds
1.6.7 Examples
df : T M −→ f ∗ T C
We look at the both sides of the above more closely. In the sequel, we set
D(f ) = f (C(f )), the set of critical values. Then, if M is compact, f defines
a C ∞ fiber bundle structure on M \ C(f ) → C \ D(f ).
We refer to [87] for a precise proof of the following
Corollary 1.6.1. In the above situation, if the critical set C(f ) of f consists
of only isolated points,
m(f, p) = (−1)m (χ(M ) − χ(F) χ(C)).
p∈C(f )
Consider first the case where the space is a complex analytic variety V ⊂ Cm
of dimension n > 1 with an isolated singularity at 0. Let U be an open
ball around 0 ∈ Cm , small enough so that every sphere in U centered at 0
meets V transversally (see [120]). For simplicity we restrict the discussion
to U . Let vrad be a continuous vector field on V \ {0} which is transverse
(outwards-pointing) to all spheres Sε around 0 for ε small enough, and scale
χ(V ) = r + χ(V ∗ ) .
2.2 Whitney Stratifications 33
Here we make a short summary of what we need in the sequel about stratifi-
cations. We refer to [28, 73, 107] for more on the subject. A stratification of a
space X is a particularly nice decomposition of this space into pieces, all of
which smooth manifolds called the strata.
li = xi yi also converges to some limiting line l, and the tangent planes Txi Vβ
converges to some limiting plane τ . The Whitney conditions (a) and (b) are
the following:
(a) The limit space τ contains the tangent space of the stratum Vα at y, i.e.,
Ty Vα ⊂ τ .
(b) The limit space τ contains all the limits of secants, i.e., l ⊂ τ .
One knows that condition (b) implies condition (a), but it is useful to have
both conditions stated explicitly.
Remark 2.2.1. Whitney stratifications are very important for several reasons,
some of which will become apparent along this text. Some important facts
about these stratifications are:
(1) Every closed (sub)analytic subset of an analytic manifold admits a Whit-
ney stratification.
(2) Whitney stratified spaces can be triangulated compatibly with the strat-
ification.
(3) The transversal intersection of two Whitney stratified spaces is a Whitney
stratified space, whose strata are the intersections of the strata of the two
spaces.
(4) Whitney stratifications are locally topologically trivial along the strata.
That is, given a complex (or real) analytic space V with a Whitney strat-
ification (Vα )α∈A , a point x ∈ Vα and a local embedding of (V, x) in Cm ,
there is a neighborhood W of x in Cm , diffeomorphic to Δ × Uα , where Uα
is a ball, neighborhood of x in Vα and Δ is a small closed disk through x of
complex dimension m − dimC Vα , transverse to all the strata of V , and such
that W ∩ Vβ = (Δ ∩ Vβ ) × Uα for each stratum Vβ with x ∈ Vβ (see [171,
§9], [5]). This is essentially a consequence of the Thom first Isotopy Lemma
(see [164]).
T W = p∗1 T Δ ⊕ p∗2 T Uα .
v = p∗1 vΔ + p∗2 vα .
The local radial extension allows to define the global radial extension. Now
we assume V to be compact.
Let us denote the corresponding filtration of V by:
where Vαj are the (not necessarily connected) strata and Vα0 is the lowest
dimensional stratum. The radial extension technique is defined by induction
on the dimension of the strata.
In the first step, let us consider an arbitrary vector field vα0 with (finitely
many) isolated singularities on the stratum Vα0 , which is compact since V is.
One performs the local radial extension (Definition 2.3.2) in a tube T (Vα0 )
around Vα0 as in (1.1.2) above, union of neighborhoods W as in Definition
2.3.2. Let us denote by v the obtained vector field. If dim Vα0 = 0, then v
is a stratified radial vector field in a ball around the point x0 ∈ Vα0 in M .
The vector field v is pointing outward T (Vα0 ) along its boundary and the
singularities of v in T (Vα0 ) are exactly those of vα0 in Vα0 . Furthermore,
Lemma 2.3.1 implies that the total Poincaré–Hopf index of v on T (Vα0 ) is
χ(Vα0 ).
The following step is to extend v around Vαj assuming the construction
being performed around Vαj−1 . That means v is already constructed in a tube
T (V αj−1 ) around V αj−1 , it is pointing outward T (V αj−1 ) along its boundary
and if x ∈ Vα ⊂ V αj−1 is a singularity of v, one has IndPH (v, x; T (V αj−1 )) =
IndPH (v, x; Vα ).
The vector field v is defined on a neighborhood of V αj \ Vαj and can be
extended as a vector field still denoted by v with (finitely many) isolated
singularities within Vαj . One considers a tube T (V αj ) around V αj as in
(1.1.2) above, union of the tube T (V αj−1 ) around V αj \ Vαj = V αj−1 and of
neighborhoods W (as in Definition 2.3.2) around Vαj . Using the local radial
extension property one extends v in T (V αj ) in such a way as induction
hypotheses are satisfied.
We may summarize the previous discussion in the following theorem of
M.-H. Schwartz (see [33] or [28] for a detailed exposition and a complete
proof):
Theorem 2.3.1. ([139, 142]) Let V be a complex analytic variety in a com-
plex manifold M , and let (Vα )α∈A be a Whitney stratification of M adapted
to V . Then there exists stratified vector fields on a neighborhood of V in M
constructed by radial extension as above. Every such vector field v satisfies:
(1) Given any stratum (Vα ), the total Poincaré–Hopf index of v on T (V α ) is
χ(V α ).
(2) v is transverse, outwards pointing, to the boundary of every small regular
neighborhood of V in M .
(3) The Poincaré–Hopf index of v at each singularity x is the same if we
regard v as a vector field on the stratum that contains x or as a vector field
2.4 The Schwartz Index on a Stratified Variety 37
The results described in the previous section tell us that given a compact
complex analytic variety V in a complex manifold M which is equipped with
a Whitney stratification {Vα } adapted to V , we may consider vector fields
on V obtained by radial extension. We now make similar considerations for
stratified vector fields in general, not necessarily obtained by radial extension.
Let us consider Kε,ε = (Bε \ Int Bε ) ∩ V . On the parts Sε = ∂Bε and
Sε = ∂Bε of the boundary of Kε,ε one has a vector field w defined by vrad and
v respectively. One extends w in Kε,ε by the radial extension process that
we described in the previous section as a stratified vector field with isolated
singularities pj so that we have
where the sum on the right runs over the singular points of w in Kε,ε . This
integer does not depend on the choice of w.
Definition 2.4.2 (Schwartz index: case of a stratified variety V and
an arbitrary vector field v with isolated singularity). The Schwartz
index of v at 0 ∈ V is defined as:
It is clear that if V is smooth at 0 then this index coincides with the usual
Poincaré–Hopf index; it also coincides with that in Sect. 1 above if 0 is an
isolated singularity of V and with the usual index of M.-H. Schwartz (2.3.3),
for vector fields obtained by radial extension.
The proof of the following theorem is exactly as that of 2.1.1 and we leave
the details to the reader.
Theorem 2.4.1. Let V be a compact, complex analytic variety in some com-
plex manifold M equipped with a Whitney stratification adapted to V . Let v be
a continuous, stratified vector field on a neighborhood of V in M , with isolated
singularities x1 , · · · , xs , all contained in V , and denote by IndSch (v, xi ; V ) the
corresponding local Schwartz indices at the singular points of V . Then one
has:
s
χ(V ) = IndSch (v, xi ; V ),
i=1
χ(V ) = 0.
Assume now we are given a vector field v defined on the regular part Vreg =
V \ Sing(V ) of V , and nonsingular away from some subcomplex S0 of Vreg .
2.4 The Schwartz Index on a Stratified Variety 39
This situation was envisaged in [30, 31] (see also [6]). In this context one
will associate an index Ind(v, S; V ) to each compact connected component S
of Sing(V ); this is the Schwartz index of v at S. This construction is relevant
for the discussion in Chaps. 8–10 of characteristic classes of singular varieties.
Let S be a connected component of Sing(V ) and T (S) a cellular tube
around S with smooth boundary ∂T (S) (Remark 1.1.2). Let us denote by
TV (S) the intersection T (S) ∩ V . The boundary ∂T (S) is transverse to V
and intersection ∂T (S) ∩ V = ∂TV (S) lies in Vreg , hence is smooth.
Let τ be a vector field tangent to Vreg , defined in a neighborhood of the
boundary ∂TV (S) in Vreg and pointing outwards TV (S) along the boundary.
Then define the Schwartz index of τ at S by:
Thus
for each λ, where dλ (v, vrad ) is the difference introduced before. Similarly, the
Poincaré–Hopf index of v in V ∗ is
Hence
the way how these vector fields extend to the ambient space. So we begin
with a brief discussion of this topic.
Let us denote by (V, 0) the germ of a complex analytic hypersurface in
Cn+1 given by a holomorphic function
which is defined on a small ball Bε and has a unique critical point at 0. Let
v be a continuous section of the bundle T Cn+1 |V . We notice that for each
x ∈ V ∗ = V \ {0}, the tangent space Tx V ∗ consists of all vectors in Tx Cn+1
which are mapped to 0 by the derivative of f :
Tx V ∗ = {ζ ∈ Tx Cn+1 dfx (ζ) = 0 }.
Hence:
= dfz (−3z 2 , 2z1 ) = 0.
dfz (ζ) 2
= f ∂f ,
df (ζ)
∂z1
∂f ∂f
ξ = 0, ,− .
∂z3 ∂z2
deg(φv ) = λ.
Alternatively one can prove that every map from a closed oriented
(2n − 1)-manifold into Wk+1 (n + k) factors by a map into the fiber
γ
S2n−1 , essentially by transversality. Hence φv represents an element
in π2n−1 Wk+1 (n + k)
Z , so φv is classified by its degree. In other words,
up to homotopy, the map φv can be regarded as a map from the link K
into the sphere S2n−1 , and deg(φv ) is its degree in the usual sense (c.f. the
following chapter where this discussion is carefully done in the real case,
which is more delicate).
Definition 3.2.1. The GSV index of v at 0 ∈ V , IndGSV (v, 0), is the degree
of the above map φv .
This index depends not only on the topology of V near 0, but also on the
way V is embedded in the ambient space. For example, the singularities in
C3 defined by
r
n
IndGSV (v, V ) = χ(V ) + (−1) μ(xi ),
i=1
Remark 3.2.2. In the above discussion we ruled out the case where the
dimension of V is 1 and V has several branches. This case is of course inter-
esting and it was first considered by Brunella [39, 40] and Khanedani–Suwa
[93] in their study of holomorphic 1-dimensional foliations on complex sur-
faces (cf. Chap. 6). In this case the GSV index is defined as the Poincaré–Hopf
index of an extension of v to a Milnor fiber. If a curve C has only one branch
at a singular point x0 this coincides with Definition 3.2.1. But if C has several
branches at x0 one has an integer attached by 3.2.1 to each branch. If C is
a plane curve, the relation among all these indices is well understood and
it is determined by the intersection number of the various branches. In fact,
Milnor in [121, Theorem 10.5 and Remark 10.10] proved that if C1 , . . . , Cr
are the irreducible components of C then one has the formula:
r
μ= μi + 2I − r + 1,
i=1
where μ, respectively
μi , is the Milnor number of C, respectively Ci , at x0
and I is defined as i<j Ci · Cj . This formula implies that if v is a vector
field on C then we have
r
IndGSV (v, x0 ; C) := IndGSV (v, x0 ; Ci ) = IndPH (v, Ft ) − 2I,
i=1
Remark 3.2.3. We notice that the definition of the GSV index works equally
well for singularities which are only geometric complete intersections [116],
not necessarily algebraic complete intersections, i.e., all we need is that the
gradient vector fields of the functions that define V are linearly independent
everywhere on V \ {0}.
∂f ∂f
df = dz1 + · · · + dz2n .
∂ z1 ∂ z2n
Assume we have chosen the coordinates in such a way that ζ has an isolated
singularity at 0, i.e., the hypersurface V = {f = 0} meets only at 0 the set
2n ∂f
i=1 { ∂zi = 0}. We set ζ = ζ|V . Notice one has:
= f ∂f
df (ζ) ,
∂z0
defined in the ambient space and which is tangent to all the nonsingular
hypersurfaces f −1 (t), t = 0. The singular set of ξ is the complete intersection
curve defined by the ideal ( ∂z∂f
1
, · · · , ∂z∂f2n ), which meets each nonsingular fiber
−1
f (t) at finitely many points, whose total sum (counting multiplicities) is
the GSV index of ζ on V . A direct computation then shows that this index,
being the intersection number of two complex varieties is equal to:
3.3 Some Applications and Examples 51
∂f ∂f
IndGSV (ζ, 0; V ) = dimC O2n+1 f, ,··· , ,
∂z1 ∂z2n
d
f (z) = z1d1 + · · · + zn+1
n+1
,
where μ(V ) is the Milnor number of V . From [121], see also Example 5.7.1
below, we know:
Hence:
Thus we conclude:
52 3 The GSV Index
F : X −→ C,
such that F −1 (0) is (isomorphic to) Y and F −1 (t) is nonsingular for all
t near 0. Assume for simplicity that X is embedded in an open subset U
of Cm . We know from [102, Th. 1.1] that for every ε >> η > 0 sufficiently
small, the restriction
F : F −1 (Sη ) ∩ Bε → Sη ,
is a fiber bundle over Sη = {z ∈ C |z| = η}. Therefore χ(Ft ), the Euler–
Poincaré characteristic of each fiber
Ft = F −1 (t) ∩ Bε , t ∈ Sη ,
is independent of t.
3.5 Nonisolated Singularities 53
It is worth noting that this index does depend on the choice of the analytic
map F chosen as a smoothing of Y and not only on Y and X. However, it is
shown in [29] that if the smoothing is given by a general linear form, then this
index determines the local Euler obstruction of X at 0 (see Chap. 8), which
depends only on Y . Hence the GSV index is independent of the smoothing if
this is given by a general linear form (this can also be proved directly). We
also notice that if (Y, 0) is a complete intersection germ then the smoothing
is essentially unique, because the base space of the universal deformation is
connected, and as we know already one has:
IndGSV (vrad , Y ; F ) = 1 + (−1)n μ,
where n is the dimension of Y and μ is its Milnor number at 0.
The results in this section are proved in [34]. Here we extend the notion of
GSV index to vector fields on complete intersection germs with nonisolated
singularities, so long as one has the strict Thom wf -condition. We begin by
recalling this condition. Then we define the index and prove the proportion-
ality theorem of [34] for this index using a geometric argument.
54 3 The GSV Index
f |f −1 (Wγ ) : f −1 (Wγ ) −→ Wγ
Definition 3.5.3. We say that the Whitney stratification (Vα )α∈A satisfies
the Thom af condition if it satisfies the above Thom condition relative to f
for each pair of strata (Vβ , Vα ) with Vα ⊂ V β .
|u, v |
δ(E, F ) = supu∈E\{0} .
v∈F \{0} uv
where d is the Euclidian distance in Cn+k and δ is the distance between linear
subspaces of Cn+k defined above.
Definition 3.5.5. We say that the Whitney stratification (Vα )α∈A of X sat-
isfies the strict Thom wf condition if it satisfies the above wf condition for
all strata.
Notice that the af condition can be expressed saying that for each sequence
xi ∈ U \ {X} converging to xo one has, if the limit exists,
lim δ Txo Vα , Txi f −1 (f (xi )) = 0.
i→∞
Hence it is clear that the wf condition implies the af condition. They are
actually equivalent in the complex analytic setting, by Teissier’s work [163].
Consider again a holomorphic map
intersects the boundary sphere ∂Bε transversely. Set T = f −1 (Dδ \ Δ). The
wf condition implies that the map
f |T : T → Dδ \ Δ
is a locally trivial fibration and by [167] we can lift vrad to vector field r in
T , whose solutions are arcs that start in ∂T = f −1 (Sδ \ Δ), Sδ = ∂Dδ , they
finish in V and they are transverse to all the “tubes” f −1 (Sη ) with η ∈]0, δ[.
This vector field vrad defines a C ∞ retraction ξ of T into V , with V as
fixed point set. The restriction of ξ to any fixed Milnor fiber F = f −1 (t0 )∩Bε ,
t0 ∈ Sδ , provides a continuous map π : F → V which is surjective and it is
C ∞ over the regular part of V . As before, we call such map ξ, or also π, a
tube map for V .
We use π to lift the stratified vector field v on V to a vector field on the
fixed Milnor fiber F. Given a point x ∈ F, we let γx be the solution of vrad
that starts at x. The end-point of γx is the point π(x) ∈ V . We parameterize
this arc γx by the interval [0, 1], with γx (0) = x and γx (1) = π(x). We assume
that this interval [0, 1] is the arc in Dδ going from to to 0, so that for each t ∈
[0, 1[ the point γx (t) is in a unique Milnor fiber Ft = f −1 (t) ∩ Bε . The family
of tangent spaces to Ft at the points γx (t) define a 1-parameter family of n-
dimensional subspaces of Cn+k that converges to an n-plane Λπ(x) ⊂ Tπ(x) (U )
when t tends to 1; one has an induced isomorphism Tx F
Λπ(x) .
Just as for hypersurfaces, since the stratification satisfies Thom’s af con-
dition, Λπ(x) contains the space Tπ(x) Vα tangent to the stratum that contains
π(x). Hence, the given vector v(π(x)) can be lifted to a vector v(x) ∈ Tx F.
Thus we obtain a vector field v, nonsingular over the inverse image of V \ {0},
which is open and dense in F, and this vector field is continuous by the wf
condition; it is also nonzero on a neighborhood of F ∩ ∂Bε since v is assumed
to have an isolated singularity at 0. Thus v has a well defined Poincaré–Hopf
index in F. As before, the homotopy class of v does not depend on the several
choices involved and we have:
q
IndGSV (v, xo ) = IndGSV (w, xi )
i=1
q
IndPH (w̃, F) = IndGSV (w, xi )
i=1
Proof. If IndPH (v, xo ) = 1 then v is homotopic to a radial vector field and the
claim follows from Proposition 3.6.1. Suppose now that IndPH (v, xo ) = −1.
Let Dxo be a small disk in Vα around xo . By [153] we can always extend v|Vα
to a vector field w on a bigger disk D in Vα containing Dxo , so that w is
transverse to the boundary of D, pointing outwards, and it has exactly three
xo , where the local index is −1 by hypothesis, and two
singular points in D:
other points x1 , x2 of local index 1. We may now construct a 1-parameter
family of vector fields on an open disk in Vα which collapses these three
singularities into a single one of index 1 at xo ; we denote the resulting vector
3.7 Geometric Applications 61
field by v. And we now extend all these vector fields by radial extension. By
Proposition 3.6.1 and Lemma 3.6.1 one has:
2
χ(F) = IndGSV (
v , xo ) = IndGSV (w, xi ) = IndGSV (v, xo ) + 2χ(F).
i=0
Hence IndGSV (v, xo )=−χ(F) and the theorem is proved when IndPH (v, xo ) =
−1. The general case follows easily: given v, we can always “morsify” its
restriction to Vα and extend the morsification by radial extension, to get a
vector field w whose singularities have all local Poincaré–Hopf indices ±1
in Vα . Thus the theorem follows from Proposition 3.6.1, Lemma 3.6.1 and
the above proof for the case of local index −1.
In this section we give applications of the GSV index to three different prob-
lems in geometry. The first gives a proof of a theorem by B. Teissier about
invariance of the Milnor number for algebraic knots; the second discusses
the triviality of the bundle that defines the canonical contact structure on
complex hypersurface germs; the third discusses the triviality of the normal
bundle on the regular part of a holomorphic foliation in the neighborhood of
an isolated singularity.
It was shown by Teissier in [162] that the Milnor number of hypersurface sin-
gularities is determined by the corresponding algebraic knot (Sε , K). That is,
Theorem 3.7.1. If two hypersurface germs (V1 , 0), (V2 , 0) in Cn+1 are such
that for sufficiently small spheres Sε1 , Sε2 the pairs (Sε1 , K1 ) and (Sε2 , K2 )
are orientation preserving homeomorphic, then μ(V1 ) = μ(V2 ).
For n > 2 this can be proved using the GSV index. For this we will use
the following Proposition:
Proposition 3.7.1. Let (Sε , K) be an algebraic knot defined by a hypersur-
face germ (V, 0). Let vrad be the restriction to K of the unit outwards normal
vector field of Sε in Cn+1 ; let τ be some (any) nowhere-zero section of the
normal bundle ν(K) of K in Sε (which is a trivial bundle). Then the degree
of the map
(vrad , τ ) : K −→ W2,n+1 ,
into the Stiefel manifold is an invariant of K, equal to
To define this contact structure, notice that the normal bundle ν(K) of
K in V has a canonical trivialization given by the unit, outwards-pointing
vector field τ of K in V , which is the restriction to K of a radial vector field
⊥
vrad on V . The complex orthogonal complement vrad of vrad at each point in
∗
V = V \ {0} is an (n − 1)-dimensional complex plane in the tangent bundle
of T V ∗ . A theorem of Varchenko establishes that the restriction of vrad ⊥
to
the link K, that we √ denote DV , determines the canonical contact structure
CV on K. If we set −1 = i as usual, then the vector field i · τ is, up to
scaling, the Reeb vector field of the contact structure.
For example, if n = 2 and the germ (V, 0) is an ICIS, then one has a
nowhere-vanishing holomorphic 2-form Ω around 0 in V ; if we equip V ∗ with
the hermitian metric induced from that in Cm , then the 2-form Ω determines
a reduction of the structure group of T (V ∗ ) from U (2) to SU (2) ∼= Sp(1), so
it determines an Sp(1)-structure on the complex bundle T (V ∗ ) (see [143]).
If, as before, we denote by τ the unit outwards normal vector field of K in V ,
then the bundle DV is the trivial 1-dimensional complex bundle spanned by
the vector field j·τ , obtained by multiplying the vector τ (x) by the quaternion
j at each point of K.
Here we give a necessary and sufficient condition for DV to be a trivial
bundle when n > 2 and the germ of V at 0 is an ICIS:
Theorem 3.7.2. The complex bundle DV that defines the canonical contact
structure on K is C ∞ trivial as a complex vector bundle if and only if the
Milnor number μ(V, 0) of the ICIS germ (V, 0) satisfies:
Lemma 3.7.2. Let W be as above and assume further that W has the
homotopy type of a bouquet of n-spheres. Let v be a continuous section of
T W |∂W which can be completed to a trivialization of the complex bundle
T W |∂W ; i.e., there exist continuous sections v2 , ..., vn of T W |∂W such that
the n-frame v (n) = {v, v2 , ..., vn } defines a trivialization of T W |∂W as a com-
plex vector bundle. Then IndPH (v, W ) is a multiple of (n − 1)!.
The proofs of these lemmas are a little technical and they are given in
detail in [146]. Here we explain only the main ideas. To motivate these, we
restrict first to the case where W is the usual 2n-ball B2n with boundary
S2n−1 . This explains where the term (n − 1)! comes from.
We recall there is a classical fibration
U (n − 1) → U (n) −→ S2n−1 ,
Dı(v(x))
ψv (x) = ,
|Dı(v(x))|
cn (W ; v (n) )[W, ∂W ] = cn (T, T]
v (n) )[T,
because v(n) extends v (n) . We claim that the latter integer is a multiple of
(n − 1)!, which obviously completes the proof of lemma 3.7.2. For this, recall
(n)
πn (U (n)) = 0 if n > 1 (see [18]), so we can assume that vo and v(n) coincide,
up to homotopy, over a parallel (S × ∗) of T, where ∗ is a point in ∂Bn .
n
Using this one may now show that the complex framing v(n) is obtained from
(n)
the trivialization vo of T W , twisting it in a neighborhood of a point, using
Kervaire’s construction (see [146] for details). So the lemma follows from the
previous discussion for the case where W is a disc.
Remark 3.7.1. Notice that if a vector field v on the ICIS (V, 0) is a component
of a trivialization v (n) of the complex bundle T V ∗ , then the GSV index of
68 3 The GSV Index
v equals (up to Lefschetz duality) the top Chern class of F relative to v (n) .
The theorem above shows that for n > 2 not all vector fields satisfy this
condition, and therefore its GSV index cannot be always expressed as a Chern
class of T F relative to a trivialization over ∂F. However, if f1 , · · · , fk are
functions defining the germ (V, 0), then (v, grad(f1 ), . . . , grad(fk )) determines
a trivialization of the bundle T Cn+k |∂F , and the corresponding relative Chern
class of degree n equals the GSV index of v. This interpretation of the GSV
index as a Chern class is closely related to the virtual index studied in Chap. 5.
Corollary 3.7.1. Let (V, 0) and F be as above. Then the normal bundle to
F in V \ {0} extends to 0 as a (continuous or smooth) vector bundle if and
only if the GSV index of v is a multiple of (n − 1)!.
An example for which this condition is not satisfied is the one of a linear
vector field on V = Cn , n > 2, since in this case the index is 1. When n = 3
one can actually say a little more:
The first corollary above is obvious from Theorem 3.7.3. We notice only
that if the normal bundle ν(F ) on V ∗ is trivial, then it is isomorphic to V ∗ ×
3.7 Geometric Applications 69
defined as in Example 3.3.1 has GSV index 0. Thus, by 3.7.3, the normal
bundle of the holomorphic foliation that v spans on V ∗ is topologically trivial.
Chapter 4
Indices of Vector Fields on Real
Analytic Varieties
Indmulti
Sch (v, 0) := (IndSch (v1 , 0), . . . , IndSch (vr , 0)).
Definition 4.1.1. Let (V, 0) be a real analytic germ as above, and let v and
w be vector fields on V that vanish only at the singular point 0 ∈ V . We say
that v and w are homotopic if there exists a continuous 1-parameter family
4.2 The GSV Index on Real Analytic Varieties 73
v → Indmulti
Sch (v, 0) := (IndSch (v1 , 0), . . . , IndSch (vr , 0)).
Proof. Let Θ(K) be the set of homotopy classes of never vanishing vector
fields tangent to V on K. By [121], V is the cone over the link K. Hence,
there is a canonical bijection between Θ(V, 0) and Θ(K). We show that Θ(K)
is classified by the Schwartz multi-index, and that for each connected com-
ponent of K there is exactly one homotopy class of tangent vector fields
corresponding to each integer, which proves the statement. It is obviously
enough to consider the case where K is connected.
Recall that given vector fields v and v on V , never-zero on K (assumed
to be connected), the difference is well defined as in Chap. 1. It is clear that
if v and v are homotopic, then their difference is zero, so they have the
same Schwartz index. Conversely, if they have the same Schwartz index, then
their difference is 0, hence they are homotopic. Thus the homotopy classes
of such vector fields are classified by their Schwartz index. It remains to see
that there are vector fields of all possible Schwartz indices, but this is easy:
let ε > ε > 0 be sufficiently small, let Kε , Kε be links of these radius and
X ⊂ V the cylinder bounded by Kε , Kε . Put on Kε a vector field v of some
given Schwartz index, say I(v); now choose in the interior of X a small disk D
and put on it a vector field v of some index I . By [153] we can extend v and
v to a vector field w on X, non singular on Kε and with no other singularity
but that on D. By construction, the Schwartz index of w is I(v) + I .
We now consider the analogous of the GSV index for vector fields on real
analytic germs. For this we first need to define the index (or degree) of a
map from a smooth (n − 1)-manifold into the Stiefel manifold Vk+1,n+k of
orthonormal (k + 1)-frames in Rn+k , with k > 0, n > 1. We recall that
Vk+1,n+k is an (n − 1)-sphere bundle over Vk,n+k and that, by [164], one
has a canonical embedding γ : Sn−1 → Vk+1,n+k , which is the fiber over
(en+1 , . . . , en+k ) ∈ Vk,n+k , where e1 , . . . , en+k is the canonical basis of Rn+k .
74 4 Indices of Vector Fields on Real Analytic Varieties
If n is even, then the (real) mod(2) GSV index of v at 0 is the integer modulo
2 defined by
IndGSV2 (v, 0) = deg2 (φv ).
In order to prove this proposition we prove first the following lemma, which
is also used later.
p
⊕ ji
r ⊕ εi
r
[N, Vk+1,n+k ] → H (N ; R) −→ ←− H0 (Ni ; R) −→
n−1 i=1 i=1
∼
H 0 (N ; R) ∼ ∼
R,
= = =
i=1 i=1
r
ζ : [N, Vk+1,n+k ] −→ R.
i=1
Proof of Proposition 4.2.1: Consider first the case n odd. From the proof
of the previous lemma we know that we have a bijection
[v] → Indmulti
GSV (v, 0) := (IndGSV (v1 , 0), . . . , IndGSV (vr , 0)).
(2) If n is even, then there is a surjection from Θ(V, 0) to ⊕ri=1 Z/2Z given
by
[v] → Indmulti
GSV2 (v, 0) := (IndGSV2 (v1 , 0), . . . , IndGSV2 (vr , 0)).
4.3 A Geometric Interpretation of the GSV Index 77
that defines the isolated complete intersection singularity (V, 0), and con-
sider the fibers f −1 (t) for t near 0. If t is a regular value of f , we call f −1 (t)
a nonsingular level surface of f . Its intersection Ft := f −1 (t) ∩ Dε with
a small disc Dε around 0 ∈ U ⊂ Rn+k is a nonsingular fiber of f . We re-
mark that we do not have in general a fibration as in the case of complex
singularities studied in [79, 116, 121]. However, by hypothesis 0 is an iso-
lated singularity in V , hence the Jacobian matrix Df (x) has rank k at each
x ∈ V \ {0}. Thus there exist ε > ε > 0 and δ > 0 sufficiently small with
respect to ε , such that Df (x) has rank k on the set Ω of all x ∈ U such that
ε > ||x|| > ε and f (x) ∈ Δδ , where Δδ is a small ball in Rk centered at 0.
By Ehresmann fibration lemma, this implies that the restriction of f to Ω is
the projection map of a locally trivial fiber bundle over Δδ . The Transversal
Isotopy Theorem [4] implies that we can move V ∩ Ω by an ambient isotopy
and take it into Ft ∩ Ω, where Ft is a nonsingular fiber of f . This carries the
vector field v to a nowhere-zero vector field on Ft ∩Ω, provided ε >> |t| > 0.
Thus one has the following lemma:
By Theorem 1.1.2 one can extend v to a vector field on the whole fiber Ft
with only one singular point, say p, in the interior of Ft . The local Poincaré–
Hopf index of v at p is independent on the way we extend v to the interior
of Ft , and this number is the Poincaré–Hopf index of v in Ft , IndPH (v, Ft ).
We note that the gradient vector fields (grad1 , . . . , gradk ) are linearly inde-
pendent on all of Ft because t is a regular value of f . Thus, if we let Dε be a
small disk in Ft centered at p and ∂Dε is its boundary, then the above map
φ : Z −→ Vk+1,n+k .
4.4 Topological Invariants and Curvatura Integra 79
p−1
χ 12 (Y ) = rankHi (Y ; Z/2Z).
i=0
(2) If n is even, then the mod(2) GSV -index of vrad equals the curvatura
integra of K.
Proof. Let vrad be as above. Up to homotopy, we can think of vrad as being the
unit outwards normal field of K in V and let (grad1 , . . . , gradk) be as before,
the gradient vector fields of (f1 , f2 , . . . , fk ). Then vrad and (grad1 , . . . , gradk )
determine a continuous map from K into the Stiefel manifold Vk+1,n+k . By
80 4 Indices of Vector Fields on Real Analytic Varieties
definition, the index of this map is the GSV -index of vrad . We recall that
cK is defined by the equality φ∗ [K] = cK [g], where φ : K → Vk+1,n+k
is the map determined by the framing. But we have that [g] = γ∗ [Sn−1 ],
where γ : Sn−1 → Vk+1,n+k is the generator of πn−1 (Vk+1,n+k ) mentioned in
Sect. 4.2. Hence φ∗ [K] = cK γ∗ [Sn−m1 ]. By 4.2.1,
and u, γ∗ [Sn−1 ] = [γ]. Under the identification of πn−1 (Vk+1,n+k ) with Z,
[γ] corresponds to 1, so that u, γ∗ [Sn−1 ] = 1. Therefore
Similarly for ind2 (φ). Moreover, the index of the corresponding map into
Vk+1,n+k does not change if we replace vrad by any other vector field on V
which is also transverse to K and is radial, outwards pointing. The result now
follows from [89, Theorem VI].
We have the following theorem:
Theorem 4.4.1. Let vrad be a radial, outwards-pointing vector field on V,
and let Ft be a nonsingular fiber of f .
(1) If n is odd, then the GSV -index of vrad equals the curvatura integra of K
and we have:
1
IndGSV (vrad , 0) = χ(K) = χ(Ft ).
2
(2) If n is even, then the mod(2) GSV -index of vrad equals the curvatura
integra of K and we have:
Proof. Statement (1) follows from the previous lemma and the fact that if X
is a compact, odd-dimensional manifold with boundary ∂X, then one always
has χ(∂X) = 2χ(X). Let us now assume that n is even. Also by the lemma
above we have that IndGSV2 (vrad , 0) is the curvatura integra of K. Thus, by
[89], Lemma 2, we have
Hence, to complete the proof of (2) we must show that χ(Ft ) = χ 12 (K) mod 2,
but this follows from 4.4.1.
The following is an immediate consequence of the theorem above.
Corollary 4.4.1. Let (V1 , 0) and (V2 , 0) be n-dimensional, isolated, complete
intersection germs in Rn+k . Let K1 and K2 be the corresponding links, and
4.5 Relation with the Milnor Number for Real Singularities 81
let F1t and F2t be the corresponding nonsingular fibers. Assume that K1 and
K2 are orientation-preserving homeomorphic. If n is odd, then
χ(F1t ) = χ(F2t ).
If n is even, then
χ(F1t ) = χ(F2t ) mod 2.
(−1)n μ = χ(F) − 1.
where vrad is the radial vector field on the isolated complete intersection
singularity (V, 0).
For real singularities a Milnor fibration does not exist in general. However,
assume
f = (f1 , . . . , fk ) : (Rn+k , 0) −→ (Rk , 0),
is an irreducible, complete intersection such that its complexification
1
ψ(K) = β(K),
2
1
ψ(K) = IndGSV (vrad , 0) = χ(K) mod 2,
2
If n is even, then:
(4) If the critical point of f is algebraically isolated, the real Milnor number
μR (f ) modulo 2 coincides with the Milnor number μ(f C ) of the complexifica-
tion.
(5) If W is a compact, oriented real analytic variety with isolated singularities
x1 , . . . , xr , which are all complete intersection germs, and if v is a continuous
vector field on W , singular at the xi s and possibly at some smooth points of
W , then the total GSV index of v is
IndGSV (v, W ) = χ(W ) + μR (xi ),
i
Proof. Statements (1), (2), and (5) are now obvious. To prove (3) we note
that by definition:
1
ψ(K) = β(K),
2
where
β(K) = dimZ2 H ∗ (K, Z2 ).
Hence,
ψ(K) ≡ β(Ft ) mod 2,
by [170]. Thus the result follows because
For statement (4) notice that by [170] we have that ψ(K) ≡ μ(f C )+ 1 mod 2,
and by definition ψ(K) ≡ μR (f ) + 1.
Chapter 5
The Virtual Index
q
c(ξ) = c(Ei )ε(i) ,
i=0
where ε(i) = (−1)i . The component of c(ξ) in H 2j (M ) is the j-th Chern class
cj (ξ) of ξ. In general, if ϕ is a symmetric polynomial, we may write ϕ(ξ) as
a polynomial in the Chern classes of ξ and express as a finite sum
(0) (q)
ϕ(ξ) = ϕk (E0 ) · · · ϕk (Eq ),
k
(i)
where, for each i and k, ϕk (Ei ) is a polynomial in the Chern classes of Ei .
Letting ∇(i) be a connection for Ei , i = 0, . . . , q, we denote by ∇• the
family of connections (∇(q) , . . . , ∇(0) ). Then ϕ(ξ) is the cohomology class of
the differential form
(0)
ϕ(∇• ) =
(q)
ϕk (∇(0) ) ∧ · · · ∧ ϕk (∇(q) ).
k
q
c(∇• ) = c(∇(i) )ε(i)
i=0
satisfying
∇(i)
A0 (M, Ei ) −−−−→ A1 (M, Ei )
⏐ ⏐
⏐
ψi
⏐1⊗ψ
i
∇(i+1)
A0 (M, Ei+1 ) −−−−→ A1 (M, Ei+1 ).
ˇ •0 ) = ϕ(∇ )
ϕ(∇
(0)
in particular ˇ = ϕ(E0 ),
ϕ(ξ)
0
where ∇ˇ • denotes the family of connections (∇(q) , . . . , ∇(1) ) for the virtual
bundle ξ̌ = qi=1 (−1)i−1 Ei . Similarly for the other “partitions” of the virtual
bundle ξ.
Note that a similar statements hold for the Bott difference form of families
of connections.
q
Let M be a C ∞ manifold, ξ = i
i=0 (−1) Ei a virtual bundle over M
and ϕ a symmetric polynomial, as before. Also let U = {U0 , U1 } be an open
covering of M . Choosing a family of connections ∇•ν = (∇ν , . . . , ∇ν ) for ξ
(q) (1)
on Uν , ν = 0, 1, we have a cochain
U
for σ = (0, σ1 , σ01 ) in A2n (U, 0 ). This again induces the integration on the
cohomology
2n
: HD (U , U0 ) → C. (5.3.4)
V
Now suppose V is compact again and let S be as above. Then the following
diagram is commutative :
∗
2n j 2n
HD (U, U0 ) −−−−→ HD (U )
⏐
⏐
⏐ ⏐ (5.3.5)
V V
=
C −−−−→ C,
Now we take connections ∇0 , ∇0 and ∇0 for T Vreg , T M |Vreg and NVreg ,
respectively, on U0 so that
(1) ∇0 is v-trivial, i.e., ∇0 (v) = 0, and that
(2) the triple (∇0 , ∇0 , ∇0 ) is compatible with (5.1.1):
Then we have
cn (∇•0 ) = cn (∇0 ) = 0
because of (1) and (2) above (cf. Propositions 5.2.1 and 1.6.1) and the
above cocycle cn (∇•∗ ) is in A2n (U, U
0 ). Hence it defines a class cn (τ, v)
2n
in HD (U, U0 ), which is sent to cn (τ ) by the canonical homomorphism
∗ 2n 2n
j : HD (U , U0 ) → HD (U ). We define the virtual index IndVir (v, S) to be the
corresponding residue. If S admits a finite number of connected components
92 5 The Virtual Index
(Sλ )λ , then we have the virtual index IndVir (v, Sλ ) for each λ. Let Rλ and
R0λ be as in Sect. 5.3. Then we may rephrase the definition of the virtual
index as (cf. [111, 149]).
Definition 5.4.1. The virtual index of v at Sλ is defined by
•
IndVir (v, Sλ ) = n
c (∇1 ) + cn (∇•0 , ∇•1 ).
Rλ R0λ
It is not difficult to show the following (cf. [156, Ch.IV, Lemma 3.3])
Lemma 5.4.1. If Sλ is in Vreg , IndVir (v, Sλ ) = IndPH (v, Sλ ).
From Theorem 5.3.7, we have the following :
Theorem 5.4.1. In the above situation, if V is compact,
IndVir (v, Sλ ) = cn (τ ).
λ V
where T f |U is the bundle of vectors tangent to the fibers of f . Starting from
v, we may construct a C ∞ vector field v on U so that it is tangent to (the
regular part of) each fiber of f , its singular set S contains C(f ) and that
the intersection of S and the fiber is compact. This is done by taking a C ∞
5.6 A Generalization of the Adjunction Formula 93
splitting of (5.5.2). Let vt denote the restriction of v to the fiber f −1 (t), for t
in Ck near 0, and St the intersection S ∩ f −1 (t), which contains the singular
set of f −1 (t). We compute IndVir (vt , St ), the sum of virtual indices of vt over
the components of St , using (the restriction to Vt of) connections as follows.
We take connections ∇0 , ∇0 and ∇0 for T f |U , T M |U and NU , respectively,
on U \ S so that
2 − 2g = −(KM + C) · C
where KM is the canonical divisor of M , g is the genus of C and the dot means
intersection of cycles. This formula follows from classical relations among
characteristic classes. In fact, 2−2g is χ(C) and this equals c1 (T C)[C], the self
intersection number C 2 is the Poincaré dual of the first Chern class of the
normal bundle of C, i.e., C 2 = c1 (NC )[C]. The canonical divisor KM is the
Poincaré dual of −c1 (T M ), essentially by definition; hence the intersection
product KM · C equals −c1 (T M |C )[C]. Therefore the Adjunction formula
follows from the exact sequence
0 −→ T C −→ T M |C −→ NC −→ 0.
This formula was generalized by Kodaira [99, 2.2] for a possibly singular
curve C in a complex surface M as:
r
= −(KM + C) · C +
χ(C) c(C, xi ), (5.6.1)
i=1
− r
xi . Since χ(C) i=1 (si − 1) = χ(C), (5.6.1) is equivalent to
r
χ(C) = −(KM + C) · C + μ(C, xi ), (5.6.2)
i=1
r+s
r
IndGSV (v, xi ) = χ(V ) + (−1)n μi .
i=1 i=1
Since the singularities of V are all isolated, Theorem 5.5.1 above says that in
the above formula, all GSV indices can be considered to be virtual indices.
Finally
n Theorem 5.4.1 tells us that the total sum of virtual indices equals
V c (τ ).
where h denotes the first Chern class of the hyperplane bundle and [ ]n the
coefficient of hn in [ ]. Thus we may compute χ(V ) from Theorem 5.6.3.
Let V0 be a nonsingular complete intersection in CPn+k
of dimension n
with the same multi-degree as V . Then we have χ(V0 ) = V0 cn (V0 ), which
is also given by the right hand side of (5.6.4). Hence we have the following
formula, which is readily proved by a direct argument as well (cf. [45, Ch.5,
(4.4) Corollary]) :
s
χ(V ) = χ(V0 ) + (−1)n+1 μ(V, pi ).
i=1
In this section, we give an integral formula for the virtual index of a holo-
morphic vector field at an isolated singular point. This is done in [111] and
is a special case of more general residue theory for holomorphic vector fields.
Since the problem is local, let U be a neighborhood of the origin 0 in
C n+k
and V a complete intersection in U defined by h1 = · · · = hk = 0,
with isolated singularity at 0. Let v be a holomorphic vector field on U ,
which is tangent to and nonsingular on V \ {0}. We may choose a coordinate
system (z1 , . . . , zn+k ) on U (see Theorem 6.3.1 below) so that, if we write as
v = n+k a
i=1 i ∂zi
∂
, the set of common zeros of a1 , . . . , an , h1 , . . . , hk consists
of only 0. Let A denote the (n + k) × (n + k) matrix whose (i, j) entry
∂zj and C the k × k matrix whose (i, j) entry cij is determined by the
is ∂a i
k
“tangency condition” v(hi ) = i=1 cij hj . Let λn = [c(A) · c(C)−1 ]n denote
the coefficient of tn in the power series expansion of
∩ V | |ai (q)| = εi , i = 1, . . . , n }
Γ = {q ∈ U
This is a special case of a more general formula for the residues of virtual
tangent bundle, see Theorem 6.3.11 below.
Abstract We have seen that for vector fields, there are indices such as the
Poincaré–Hopf index and the virtual index, that arise from localizations of
certain Chern classes. If the vector field is holomorphic, the localization the-
ory becomes richer because of the Bott vanishing theorem, and this produces
further interesting residues. This theory can be developed for general sin-
gular foliations on certain singular varieties. We consider here the case of
holomorphic vector fields and the slightly more general case of one dimen-
sional singular foliations. We refer to [156] for a systematical treatment of
the general case.
Here we have three types of residues:
(1) Baum–Bott residues and generalizations to singular varieties,
(2) Camacho–Sad index and various generalizations,
(3) Variations and generalizations.
In all the above cases the residues arise from a Bott type vanishing theo-
rem, which in turn comes from an action of the vector field or the foliation
on some vector bundle or virtual bundle. The residues of type (1) were first
introduced by R. Baum and P. Bott in [13,14]. In general these arise from the
action of the foliation on the normal sheaf of the foliation. The Camacho–Sad
index (2) was introduced in [42] and was effectively used to prove the exis-
tence of a separatrix at a singular point of a holomorphic vector field on the
complex plane. Nowadays there are many generalizations of this index, see
Remark 6.3.3 below. These residues arise from the action of the foliation on
the normal bundle of an invariant subvariety. The residues of type (3) were
first introduced by B. Khanedani and T. Suwa in [93] and generalized in
[113]; see also the related articles [39] and [40] by M. Brunella. These type of
residues arise from the action of the foliation on the ambient tangent bundle.
These three types of residues are listed above in historical order, but they
are explained below in the reversed order, for logical reasons.
In each case, the residue at an isolated singularity can be expressed in
terms of a Grothendieck residue on singular variety.
S(v) = { p ∈ M | v(p) = 0}
αv : A0 (M0 , E) −→ A0 (M0 , E)
Note that a similar result holds for the difference form of a finite number
of connections and for families of v-connections of virtual bundles.
We see that T M0 becomes a holomorphic v-bundle by the action
αv : A0 (M0 , T M0 ) −→ A0 (M0 , T M0 )
given by αv (w) = [v, w]. Note that there exists a v-connection which is
v-trivial.
We will see that, for a homogeneous symmetric polynomial ϕ of degree m,
the class ϕ(T M ) is localized at S(v). Here the vector field v is the “geometric
object” of Sect. 1.6.2, a “special connection” is a v-connection and the relevant
“vanishing theorem” is Theorem 6.1.1.
6.1 Baum–Bott Residues of Holomorphic Vector Fields 99
Set S = S(v). Then from the arguments in Sect. 1.6.2, we have a class in
H 2m (M, M \ S; C), which we denote by ϕS (T M, v) and call the localization
of ϕ(T M ) by v at S. It is sent to ϕ(T M ) by the canonical homomorphism
j ∗ : H 2m (M, M \ S; C) → H 2m (M, C).
Definition 6.1.3. Suppose S = S(v) is compact. The Baum–Bott residue
Resϕ (v, T M ; S) (sometimes abbreviated as Resϕ (v, S)) is the image of the
class ϕS (T M, v) by the Alexander isomorphism
∼
H 2m (M, M \ S; C) −→ H0 (S, C).
The residue Resϕ (v, T M ; S) is a complex number given by the right hand
side of (1.6.4).
If S has only a finite number of connected components (Sλ )λ , we have the
residue Resϕ (v, T M ; Sλ ) for each λ.
The above residues are originally introduced in [13, 14], where they are
defined using forms with compact support.
From the above argument and Theorem 1.6.5, we have the following
theorem.
Theorem 6.1.2. In the above situation,
(1) For each connected component Sλ of S(v), we have a well-defined residue
Resϕ (v, T M ; Sλ ).
(2) If M is compact,
Resϕ (v, T M ; Sλ ) = ϕ(T M ).
λ M
Remark 6.1.2. 1. The above definition of the matrix ϕ(H) differs from that
in [156] by a normalization constant, i.e., in [156], it is defined as ϕ(H) =
√ i
P (c1 (H), . . . , cm (H)), ci (H) = −1/2π σi (H).
√ m
2. In particular, if ϕ = cm = −1/2π σm , noting that
we have
da1 ∧ · · · ∧ dam
Rescm (v, 0) = Res0 ,
a1 , . . . , am
which represents the Poincaré–Hopf index IndPH (v, 0) of v at 0 (cf. Example
1.6.2).
m
∂
v= λi zi , λi ∈ C.
i=1
∂zi
Then, since A in this case is the diagonal matrix with diagonal entries
λ1 , . . . , λm , the residue with respect to a polynomial ϕ of degree m is given by
ϕ(λ1 , . . . , λm )
Resϕ (v, 0) = .
λ1 · · · λm
Thus, in general, the Baum–Bott residues are not necessarily integers (and
even not real numbers).
6.2 One-Dimensional Singular Foliations 101
vβ = fαβ vα on Uα ∩ Uβ . (6.2.1)
ι : F −→ T M,
α(f v, s) = f αv (s).
defined by α(v, η(w)) = η([v, w]) is well-defined and makes NF0 an F0 -bundle.
Let νF = T M − F be the virtual normal bundle of the foliation F . We
will see that, for a homogeneous symmetric polynomial ϕ of degree m, the
characteristic class ϕ(νF ) is localized at S(F ).
Let S = S(F ), U0 = M \ S and U1 an open neighborhood of S in M . Let
∇•0 = (∇0 , ∇0 ) be a pair of connections for F and T M , respectively, on U0
and ∇•1 = (∇1 , ∇1 ) one for F and T M on U1 . Then ϕ(νF ) is represented by
the cocycle
ϕ(∇•∗ ) = (ϕ(∇•0 ), ϕ(∇•1 ), ϕ(∇•0 , ∇•1 ))
in A2m (U). Now let ∇ be an F0 -connection for NF0 on U0 and choose (∇0 , ∇0 )
so that the triple (∇0 , ∇0 , ∇) is compatible with the sequence (6.2.2) on U0 .
Then, by Proposition 5.2.1 and Theorem 6.2.3,
6.2 One-Dimensional Singular Foliations 103
ϕ(∇•0 ) = ϕ(∇) = 0
and thus ϕ(∇•∗ ) is in A2m (U, U0 ) and it defines a class in H 2m (M, M \ S; C),
which we denote by ϕS (νF , F ) and call it localization of ϕ(νF ) by the foliation
F at S.
If S is compact, as in Sect. 1.6.2, we have the residue Resϕ (F , νF ; S) (also
called the Baum–Bott residue). We sometimes abbreviate it as Resϕ (F , S).
Moreover, if S has a finite number of connected components (Sλ )λ , we have
the residue Resϕ (F , νF ; Sλ ) for each λ.
Let Rλ be a real 2m-dimensional manifold with C ∞ boundary in U1 con-
taining Sλ in its interior and disjoint from the other components. Let also
R0λ = −∂Rλ . Then the residue Resϕ (F , νF ; S) is a complex number given by
Resϕ (F , νF ; Sλ ) = ϕ(∇•1 ) + ϕ(∇•0 , ∇•1 ). (6.2.4)
Rλ R0λ
The above residues are also originally introduced in [13, 14], where the
residues are defined using forms with compact support.
From the above argument and Theorem 1.6.5, we have the following
theorem.
where the right hand side is the residue in Sect. 6.1. In particular, if
Sλ consists of a point p, then Resϕ (F , νF ; p) is given by the formula in
Theorem 6.1.3.
If F is generated by a global vector field, then F is trivial and the results
reduce to those in Sect. 6.1
2. Note that, in general, the bundle T M0 or F0 does not admit F0 -actions
and that the bundle NF0 does not extend to a bundle on M . Thus the residue
Resϕ (F , νF ; S) is not of the type considered in the first part of Sect. 6.3.4 or
Sect. 6.3.3 below.
104 6 The Case of Holomorphic Vector Fields
3. The bundle map F → T M induces an injective map on the sheaf level with
quotient NF , which is the normal sheaf of the foliation. We have ϕ(νF ) =
ϕ(NF ). The residue theory for general singular foliation is developed using
the sheaf NF ([14, 156]).
4. See [147] for the case of manifolds with boundary.
∩ V | |fi (q)| = εi , i = 1, . . . , n }
Γ = {q ∈ U
n+k
∂
v = ai (z1 , . . . , zn+k )
i=1
∂zi
αv (w) = [ V0 .
v , w]| (6.3.2)
Resϕ (F , νF |V ; S) = Resϕ (
v , T M |V ; S).
6.3 Residues of Holomorphic Vector Fields on Singular Varieties 107
The residues in this subsection are introduced in [112], generalizing the ones
already known in various special cases (cf. Remark 6.3.3 below).
In this section, we assume that V is an LCI of dimension n defined by a
section of a holomorphic vector bundle N over M . Recall that N |Vreg = NVreg ,
the normal bundle of the regular part Vreg . We set NV = N |V . Let v be
a holomorphic vector field on a neighborhood of V in M which leaves V
invariant. Letting v be the vector field on (the regular part of) V induced by
v,
we define the singular set S = S(v, V ) as before. Then NV0 , V0 = V \ S(v, V ),
is a v-bundle with the action
V0 ) with w
For ν ∈ A0 (V0 , NV0 ), we may write ν = π(w| a section of T M
in a neighborhood of V0 . Then we set
αv (ν) = π([ V0 ).
v , w]| (6.3.6)
Then it does not depend on the choice of w and defines a holomorphic action.
For a symmetric homogeneous polynomial ϕ of degree n, we have the local-
ization of ϕ(NV ) at S, which we denote by ϕ(NV , v ), instead of ϕ(NV , v),
since it depends on the extension v of v. Moreover, if S is compact, we have
the residue Resϕ (
v , NV ; S). From Theorem 5.3.7, we have the following
k
v(fi ) = kij fj ,
i=1
for some holomorphic functions kij ([136], [38]). Let K = (kij ) and ϕ a
homogeneous symmetric polynomial of degree n. Then we have the following
formula (cf. [156, Ch. IV, Theorem 6.3]).
Resϕ (F , NV ; S) = Resϕ (
v , NV ; S).
A1 = ∂(a1 , . . . , an )/∂(z1 , . . . , zn )
A2 = ∂(an+1 , . . . , an+k )/∂(zn+1 , . . . , zn+k ).
[2, 21, 24]. Furthermore, in [3] the residue theories for holomorphic maps and
for foliations are unified and much generalized using a version of vanishing
theorem finer than the one we expect to have by assuming the involutiveness.
In [22,23], the residue theory for fixed subvarieties of holomorphic self-map
of certain singular varieties is developed. It is used to prove the existence of
a parabolic curve at a fixed point of a holomorphic self-map of surfaces with
certain type of singularity.
In fact, for this residue, we only need a vector field on the nonsingular
part of V . Thus, suppose we are given a holomorphic vector field v on the
regular part of V . We denote by S = S(v, V ) the union of Sing(V ) and the
singular set S(v) of v and set V0 = V \ S. Let U , U0 and U 1 be as before. On
V0 , T V0 is a v-bundle with the action given in Sect. 6.1. This action defines
a localization ϕ(τV , v) of ϕ(τV ) at S. If S is compact, we have the residue
Resϕ (v, τV ; Sλ ) for each connected component Sλ of S.
Note that, if we have an extension v of v, we have
Resϕ (
v , τV ; Sλ ) = Resϕ (v, τV ; Sλ ).
From Proposition 6.3.1, (6.3.13) and Remark 6.3.5, 1, we have the following:
Proposition 6.3.2. Let M , V and F be as above and let S be a compact
connected component of S(F , V ). Suppose there is a vector field v defining F
6.3 Residues of Holomorphic Vector Fields on Singular Varieties 113
In view of the above, we denote IndGSV (v, p) by IndGSV (FV , p). The fol-
lowing theorem is a consequence of Theorems 5.5.1, 5.6.3 and 6.3.14 and
Proposition 6.3.1.
r
IndGSV (FV , pi ) = cn (τ − F ).
i=1 V
There are some interesting relations among the residues in Sects. 6.3.2–
6.3.4. A general formula is given in [156, Ch.IV, 7]. Here we give a formula
in the case n = k = 1. Thus V = C is a possibly singular curve in (a
nonsingular) surface M . In this case, we have
There are explicit examples of the residues in [156, Ch.IV, 8]. Example 5.7.1
is one of them.
Chapter 7
The Homological Index and Algebraic
Formulas
vector fields on V with isolated singularities, defined in the usual way. This
total index is independent of the choice of vector field, being therefore an
invariant of V . If V is a local complete intersection as in Sect. 5.4, then the
corresponding global invariant is the 0-degree Fulton–Johnson class of V (see
in Chap. 11).
It would be interesting to know what the homological index measures for
singular germs and varieties which are not local ICIS. This is related with
extending the notion of Milnor number to isolated singularity germs which
are not complete intersections (see Chap. 9).
The basic references for this section are the articles by Gómez-Mont and
various co-authors, see [17, 62–65, 68–70]. See also [7, 92, 97].
An algebraic formula for the index of holomorphic vector fields on singular
varieties was given in [71], inspired by (7.0.1), but that formula applies only
under very stringent conditions: for holomorphic vector fields on a hypersur-
face germ V which are tangent to the fibers of a defining function f of V .
Using the fact that in the classical case, when the ambient space is smooth,
the Poincaré–Hopf local index can be interpreted as the Euler-characteristic
of a certain Koszul complex (see [75]), Gómez-Mont introduced in [68] a
notion of the homological index of holomorphic vector fields. Let us explain
this invariant.
Let (V, 0) ⊂ (Cm , 0) be a germ of a complex analytic variety of pure
dimension n, which is regular on V \ {0}. So V is either regular at 0 or else
it has an isolated singular point at the origin. A vector field v on (V, 0) can
always be defined as the restriction to V of a vector field v in the ambient
space which is tangent to V \ {0}; v is holomorphic if v can be chosen to
be holomorphic. So we may write v as v = (a1 , · · · , am ) where the ai are
restriction to V of holomorphic functions on a neighborhood of 0 in (Cm , 0).
It is worth noting that given any space V as above, there are always holo-
morphic vector fields on V with an isolated singularity at 0. This nontrivial
fact is indeed a weak form of a stronger result ([16, p. 19]): in the space
Θ(V, 0) of germs of holomorphic vector fields on V at 0, those having an iso-
lated singularity form a connected, dense open subset Θ0 (V, 0). Essentially
the same result implies also that every v ∈ Θ0 (V, 0) can be extended to a
germ of holomorphic vector field in Cm with an isolated singularity, though
it can also be extended with a singular locus of dimension more that 0, a
fact that may be useful for explicit computations (c.f. the last part of the
following section).
A (germ of) holomorphic j-form on V at 0 means the restriction to V
of a holomorphic j-form on a neighborhood of 0 in Cm ; two such forms in
Cm are equivalent if their restrictions to V coincide on a neighborhood of
7.1 The Homological Index 117
j
0 ∈ V . We denote by ΩV,0 the space of all such forms (germs); these are the
0
Kähler differential forms on V at 0. So ΩV,0 is the local structure ring O(V,0)
j 0
of holomorphic functions on V at 0 and each ΩV,0 is an ΩV,0 -module. Notice
that if the germ of V at 0 is determined by (f1 , · · · , fk ) then one has:
j
ΩV,0 := ΩCj m ,0 /(f1 ΩCj m ,0 + df1 ∧ ΩCj−1 j j−1
m ,0 , . . . , fk ΩCm ,0 + dfk ∧ ΩCm ,0 ), (7.1.1)
0 −→ ΩV,0
n
−→ ΩV,0
n−1
−→ · · · −→ OV,0 −→ 0, (7.1.2)
In particular the complex is exact if v(0) = 0. Since the contraction maps are
OV,0 -module maps, this implies that if V has an isolated singularity at the
origin, then the homology groups of this complex are concentrated at 0, and
they are finite dimensional because the sheaves of Kähler differentials on V
are coherent. Hence it makes sense to define, for V a complex analytic germ
with an isolated singularity at 0 and v a holomorphic vector field on V with
an isolated singularity at 0:
n
•
Indhom (v, 0; V ) = (−1)i hi (ΩV,0 , v),
i=0
118 7 The Homological Index and Algebraic Formulas
•
where hi (ΩV,0 , v) is the dimension of the corresponding homology group as
a vector space over C.
where the sum on the right runs over the singularities of v at regular points
of V near 0.
f (x, y) = xk + y k , k > 1,
then the radial vector field v = (x, y) is obviously tangent to V , singular only
at 0, and a straight-forward computation using the formulas in Sect. 2 below
shows that its GSV index (which equals the homological one) is 1 − (k − 1)2 ,
which is negative if k > 2. However the results in [16] show that in all cases
there is a smallest possible index that the holomorphic vector fields on an
isolated singularity germ (V, 0) may attain, and this index is attained by an
open and dense set in the space of germs of holomorphic vector fields at 0;
what is this lower bound?
Obviously this provides an invariant of the singularity which is interesting
to study. This is closely related to asking what is a generic vector field on a
singular variety?
Conjecture: Let (V, 0) be an ICIS germ of dimension n ≥ 1, with Milnor
number μ ≥ 1, and let v be a holomorphic vector field on V with an isolated
singularity at 0. Then the Poincaré–Hopf index of any continuous extension
of v to a nearby Milnor fiber of V is ≥ 1 + (−1)n μ.
It is not hard to show that this conjecture is true for curves, i.e., when V
has complex dimension 1, and for quasi-homogeneous germs of all dimensions.
algebraic facts in [68] that lead to these statements is not an easy task and
requires hard computations, so we only sketch here some of the main points.
We refer to Gómez-Mont’s article for the actual proofs and details. We refer
to [17] for the equivalent arguments for the index of vector fields on complete
intersection singularities.
Let B = Bn+1 be an open ball around the origin in Cn+1 and let f :
B n+1
→ C be a holomorphic n+1 function with 0 ∈ Cn+1 as its only critical
point. Hence the 1-form i=1 dzi dzi vanishes only at 0 ∈ Cn+1 , where dz
∂f ∂f
i
denotes the partial derivative of f with respect to zi . Let If be the Jacobian
∂f
ideal ( dz 1
, · · · , dz∂f
n+1
) ⊂ OB,0 of f . Given a vector field v tangent to V , with
a unique singularity at 0, restriction of a vector field v = (a1 , · · · , an+1 ) on
B with a unique singularity at the origin, it is shown in [68, Theorem 2] that
•
the homology groups Hi (ΩV,0 , v) of the complex (7.1.2) have dimensions:
•
h0 (ΩV,0 , v) = dimC OB,0 /(f, a1 , · · · , an+1 ),
•
hn (ΩV,0 , v) = dimC OB,0 /(f, If ),
•
hi (ΩV,0 , v) = λ, for i = 1, . . . , n − 1,
noticing that the tangency condition for v means that v(df ) is a multiple
of f , so that df v ) is a holomorphic function on B. As a consequence of
f (
these computations Gómez-Mont deduced the following expressions for the
homological index (Theorem 1 in [68]).
The next step is easy once we have Theorem 7.1.3 above, i.e., the stability
of the index under perturbations. In fact, one has:
is a constant that depends only on K and not on the choice of vector field.
defined on all of C2n and with 0 as its only singularity. The differential of
f is:
∂f ∂f
df (z1 , · · · , z2n ) = (z1 , · · · , z2n ) dz1 + · · · + (z1 , · · · , z2n ) dz2n .
∂z1 ∂z2n
everywhere on C2n \ {0}. This means that ζ is not only tangent to the hyper-
surface V := f −1 (0), but it is also tangent to all the nonsingular hypersurfaces
Vt := f −1 (t), and with no singularity on each Vt , t = 0. This implies that its
GSV index is 0. On the other hand, in this case Theorem 7.2.1 says that the
homological index is:
since in this case the components of ζ are the partial derivatives of f , so they
generate the Jacobian ideal If . Hence both indices coincide for vector fields
on hypersurface singularities of odd dimension.
When the dimension of V is even the considerations are similar in spirit:
one constructs a holomorphic vector field on every such singularity germ,
whose homological and GSV indices can be computed explicitly and to show
that they coincide. However the construction of such examples is not that
simple now. Let us denote the coordinates of C2n+1 by (z0 , z1 , · · · , z2n ) and
consider the vector field:
∂f ∂f ∂f ∂f
v = f, ,− ,··· , ,− ,
∂z2 ∂z1 ∂z2n ∂z2n−1
df ∂f
(
v) = ,
f ∂z0
It is natural to ask whether one can also find an algebraic formula for comput-
ing the index of a C ∞ , or real analytic, vector field at an isolated singularity.
It seems that this question was first raised by V. I. Arnold (c.f. [172]), and the
answer was given independently by D. Eisenbud and H. Levine in [58] and by
G.N. Khimshiashvili in [94], proving that the index can be computed as the
signature of an appropriate bilinear form. This signature formula is proved
124 7 The Homological Index and Algebraic Formulas
The formula of [58, 94] is somehow the paradigm of all the later algebraic
formulas for indices of analytic vector fields on real analytic singular varieties.
Let us denote by ARn ,0 the local ring of germs of real analytic real-valued
functions, and consider a germ of a vector field v = (a1 , · · · , an ) at 0, where
the components are elements in ARn ,0 . We let Bv be the local algebra of v:
Bv := ARn ,0 /(a1 , · · · , an ),
For simplicity we also denote by Jv the residue class of the Jacobian in the
local ring Bv . Now observe that Bv is actually an algebra and given a linear
f,g
functional φ : Bv → R one can define a map Bv × Bv −→ R by:
· φ
Bv × Bv −→ Bv −→ R.
This is clearly a bilinear form. Let Sgn(v) denote the signature of this bilinear
form, i.e., the number of positive eigenvalues minus the number of negative
eigenvalues. Then one has the index formula of [58, 94]:
Theorem 7.3.1. One can always choose the linear form φ so that φ(Jv ) > 0,
and in this case one has:
Remark 7.3.1. If the real analytic vector field has an isolated singularity that
is not algebraically isolated, the signature formula for the index does not hold
for several reasons. In [43] V. Castellanos used the Koszul complex analogous
to (7.1.2) with real analytic forms replacing the holomorphic ones, to obtain
a signature formula in this setting.
126 7 The Homological Index and Algebraic Formulas
We know that for real analytic hypersurfaces there is not “a” Milnor fiber
but there are fibers to the left and to the right of a critical value, say 0 ∈ R,
with possibly different topology. Therefore one can not define in general an
index in the spirit of the GSV-index, i.e., a well-defined integer associated to
each vector field with an isolated singularity, which measures the number of
zeroes of an extension of the vector field to a Milnor fiber: the number one
gets depends on the choice of Milnor fiber (see [6] or Chap. 4 above).
If the hypersurface is odd dimensional, things are simpler from the topolog-
ical viewpoint because the Euler–Poincaré characteristic of the Milnor fiber
is well defined; however for even dimensions this is only well defined mod-
ulo 2. Still, the formulas by Arnold in [9, p. 3] show that for gradient vector
fields, the algebra behind the function determines the Euler characteristic of
the fibers in all cases: it is 1 ± σ where σ is the signature of an appropriate
bilinear form on the local ring (algebra) of the singularity.
This suggests that something can be done for vector fields in general on
real analytic hypersurface singularities. This program was carried out in-
dependently (and differently) by Gómez-Mont and Mardešić on one hand
([69, 70], see also [62, 66]) and by Ebeling and Gusein-Zade [49] on the other
hand.
The work in [49] is somehow inspired by the formulas in [78, 166]; for this
the authors define an index of vector fields with an isolated singularity, which
is the radial index that we introduced in Chaps. 2 and 4, and then they give an
algebraic formula to compute this index when the vector field is the gradient
of a function.
The work of Gómez-Mont and Mardešić is closely related to the formula in
[58,94]; let us have a glance of what they do. Let U be an open neighborhood
around 0 in Rn+1 , and let f : (U, 0) → (R, 0) be analytic with an algebraically
isolated singularity at 0 ( i.e., its complexification has an isolated singularity);
set V = f −1 (0). Let A be the local ring of f at 0 (an algebra in fact):
∂f ∂f
A = ARn+1 ,0 /( ,··· , ),
∂x1 ∂xn+1
∂f
where ( ∂x 1
, · · · , ∂x∂f
n+1
) = Jf is the Jacobian ideal of f . This algebra is finite
dimensional because of the assumption that V has an algebraically isolated
singularity, and it has a distinguished element: the class of the Hessian,
∂ 2 f
Hess(f ) := det ∈ A.
∂xi ∂xj
Jv
Jf (v) := ∈ Bv /AnnBv (hv ),
hv
where AnnBv (hv ) is the annihaltor, hv being as above. It is proved that there
is a linear map : Bv /Ann(h) → R such that (Jf (v)) > 0. The product
in Bv /Ann(h) together with defines a bilinear form on Bv /Ann(h). Let
SgnV,0 (v) denote the signature of this bilinear form.
It is proved in [69] that the function SgnV,0 “behaves like an index” in the
sense that for n even it satisfies the law of conservation of number:
SgnV,0 (v) = SgnV,0 (vt ) + IndexPH (vt , x; V \ {0})
x∈V \{0}
vt (x)=0
for x close to 0 and vt tangent to V and close to v. The same formula holds
for n odd under a certain additional hypothesis.
Similarly, the relative Hessian is defined in [70] by:
Hess(f )
Hessrel (f ) := ∈ A/AnnA (hv ).
hv
128 7 The Homological Index and Algebraic Formulas
where hv = df (v)/f ∈ A.
We refer to [70] for the proof of this result and for several explanations
giving insights of the geometry and algebra behind this formula. Notice that
if V is regular at 0, then this formula reduces to the one in [58, 94]. We refer
to [69] for a discussion when V is even-dimensional.
Chapter 8
The Local Euler Obstruction
is a map σ : Vreg → U × G(n, m) defined by σ(x) = (x, Tx (Vreg )). The Nash
transformation V of V is the closure of Im(σ) in U × G(n, m). It is a (usually
singular) complex analytic space endowed with an analytic projection map
ν : V −→ V
T −→ V −→ V.
π ν
Remark 8.1.1. We notice that for ε > 0 small enough, if v is a nowhere zero
section of T defined on ν −1 (V ∩ Sε ) which lifts a vector field transverse to
V ∩Sε , then the Euler obstruction EuV (0) equals the obstruction to extending
v as a nowhere zero section of T over ν −1 (V ∩ Bε ). This is a consequence of
the fact that every section v as above is homotopic to a section of T over
ν −1 (V ∩ Sε ) obtained by lifting a radial vector field of V at 0. Hence, to
calculate the Euler obstruction of (V, 0) it suffices to construct a nowhere
zero section of T defined on ν −1 (V ∩ Sε ) which lifts a vector field transverse
to V ∩ Sε . Of course one also needs to understand how this section extends
to ν −1 (V ∩ Bε ).
The following result summarizes some basic properties of the Euler ob-
struction:
Theorem 8.1.1. The Euler obstruction satisfies:
(1) EuV (x) = 1 if x is a regular point of V .
(2) EuV ×V (x × x ) = EuV (x) · EuV (x ).
(3) If V is
locally reducible at x and Vi are its irreducible components, then
EuV (x) = EuVi (x).
(4) EuV (x) is a constructible function on V , in fact it is constant along the
strata of a Whitney stratification.
These statements are all contained in [117], except for (iv) which is im-
plicitly stated there and we refer to [33] for a detailed proof. Now we have
the following result of [33].
In this section, we prove the Proportionality Theorem for vector fields in [33]:
Theorem 8.1.2. Let v be a stratified vector field on V which is obtained
by radial extension in a neighborhood of a singularity x ∈ Vα . Let v be the
canonical lifting of v to a section of the Nash bundle T over the boundary
of ν −1
(V ∩ Bε (x)), where Bε (x) is asmall ball around x in Cm . Let O(v) ∈
H ν (V ∩ Bε (x)), ν −1 (V ∩ Sε (x)) be the obstruction cocycle to extending
2n −1
In short this theorem says that the obstruction EuV (v, x) to extend the
lifting v as a section of the Nash bundle inside ν −1 (V ∩ Bε (x)) is propor-
tional to the Poincaré–Hopf index of v at x, the proportionality factor being
precisely the local Euler obstruction.
Let p be a point in a stratum Vα with nα = dimC Vα > 0. Let B a ball
around p in M , small enough so that T Vα is trivial on B∩Vα , and set S = ∂B.
We denote by T × M , T × Vα and T× the bundles obtained from T M , T Vα and
T, respectively, by removing the zero sections.
We denote by θ(M, V ; p) the set of homotopy classes of stratified vector
fields on V obtained by radial extension of some vector field around p on Vα .
We remark that, by [48] Theorem 1.1, this set coincides with the set of usual
homotopy classes of stratified vector fields.
On the other hand, we denote by θ(Vα , p) the set of (usual) homotopy
classes of vector fields of Vα , defined and nonvanishing on S ∩ Vα . Note that
Sα := S ∩ Vα is a (2nα − 1)-sphere, so that such a vector field vα defines a
map
pr2
T × Vα |Sα −→ Sα × Cnα \ {0} −→ Cnα \ {0},
vα h
ϕvα : Sα −→
where h is an isomorphism.
This correspondence induces a bijection of θ(Vα , p) onto the homotopy
group π2nα −1 (Cnα \{0})
Z, where the isomorphism is given by the mapping
degree. A vector field vα as above may be extended to a vector field on B
with isolated singularity at p. Recall that, by definition, IndPH (vα , Vα ; p) is
the mapping degree of ϕvα . In summary, θ(Vα , p)
Z, which is generated by
the class [vα,rad ] of a radial vector field on Vα , singular at p.
Lemma 8.1.2. There is a natural bijection between θ(M, V ; p) and θ(Vα , p).
Thus
θ(M, V ; p)
Z.
It is generated by the class [vrad ] of a stratified radial vector field. Moreover,
the elements in θ(M, V ; p) are classified by their local Schwartz index at p.
Proof. Note that the map
θ(M, V ; p) −→ θ(Vα , p)
ϕ ◦ κ : Sα −→ Cnα \ {0}.
8.2 Euler Obstruction and Hyperplane Sections 133
ψ : (S ∩ V ) × [0, 1] −→ T × M |S∩V
between v and k · vrad , k = IndSch (v, V ; p). Here k · vrad denotes the radial
extension of the vector field k · vα,rad on S ∩ V . Thus we have
∂ Im ψ = v(S ∩ V ) − k · vrad (S ∩ V )
and we have
∂ Im ψ = v(ν −1 (S ∩ V )) − (k
· vrad )(ν −1 (S ∩ V ))
(k
· vrad )(ν −1 (S ∩ V )) = k · v&
rad (ν
−1
(S ∩ V ))
The idea of studying the Euler obstruction “à la” Lefschetz, using hyperplane
sections, is found in the work of Dubson [46] and Kato [88]. Also in [106] there
are results in this spirit for the Euler obstruction and also for the Chern
134 8 The Local Euler Obstruction
classes of singular varieties. The approach we follow here is that of [29, 32],
which is topological.
We start with the following lemma, which is a special case of well-known
results about Lefschetz pencils. Let us denote by L the space of complex linear
forms on Cm . Fix a Whitney stratification of V . There are a finite number of
strata of this Whitney stratification which contain 0 in their closure, and we
assume that the representative of (V, 0) is chosen small enough so that these
are the only strata of V .
Lemma 8.2.1. [29] There exists a nonempty Zariski open set Ω in L such
that for every l ∈ Ω, there exists a representative V of (V, 0) so that:
(1) for each x ∈ V , the hyperplane l−1 (0) is transverse in Cm to every limit
of tangent spaces in T Vreg of points in Vreg converging to x,
(2) for each y in the closure V α in V of each strata Vα , α = 1, . . . , , the
hyperplane l−1 (0) is transverse in Cm to every limit of tangent spaces in T Vα
of points converging to y.
In particular, for each l ∈ Ω one has for the Nash transformation
V ⊂ Cm × (G(n, m) \ H ∗ ),
EuV (0) = χ(Vα ∩ Bε ∩ l−1 (t0 )) · EuV (Vα ),
α=1
where χ denotes the Euler–Poincaré characteristic and EuV (Vα ) is the value
of the Euler obstruction of V at any point of Vα , α = 1, . . . , .
Remark 8.2.1. As noticed in [32], Theorem 8.2.1 can be stated through the
framework of bivariant theory [26, 61] : the local Euler obstruction, as a con-
structible function, satisfies the local Euler condition with respect to general
linear forms.
8.2 Euler Obstruction and Hyperplane Sections 135
and of the fact that the local Euler obstruction at a nonsingular point is equal
to 1. Observe that, since each xi is an isolated singular point of Ft , we can
apply 8.2.1 to compute the right hand side of 8.2.3, because
Now we have the following result of [32]; this compares the Euler obstruction
of the space V with that of a function on V . According to Proposition 8.5.1,
Theorem 8.2.1 is a special case of Theorem 8.4.1 taking f to be a general
linear form.
For the proof of Lemma 8.4.1, we refer the reader to [32]. The first steps
of the proof are an interesting application of M.H. Schwartz techniques in
order to construct the vector field w on a tube Bε ∩ f −1 (Dδ ) \ {0}, transverse
(outwards pointing) to the boundary of the tube. The final step is to extend
this vector field to all of V using Theorem 2.3 in [29]. Let us show how one
deduces Theorem 8.4.1 from Lemma 8.4.1 (see [32]):
ν −1 (Vε )) = Obs(w,
Obs(w, ν −1 (Vε )) + Obs(w,
ν −1 (Vε ,ε ))
Euf,V (0) = 0.
140 8 The Local Euler Obstruction
The proof of this Proposition is implicit within the proof of 2.3 in [29] and
is also a consequence of Theorems 8.2.1 and 8.4.1 together. However we prove
it here, following [32], for completeness and because this is how one deduces
formula 8.2.1 from 8.4.1.
Proof. In a first step define the map
F
T ⊂ (U × G(n, m)) × Cm −→ Dδ ⊂ C
This section is taken from [150], by J. Seade, M. Tibăr and A. Verjovsky. Here
we show how stratified Morse theory yields to a clear understanding of what
the invariant Euf,V (x) is for arbitrary functions with an isolated singularity.
These results can also be deduced from Schürmann’s book [138], and also
from the work of D. Massey, [118, 119]. For this we recall the definition of
complex stratified Morse singularities (see Goresky–MacPherson [73], p. 52).
Definition 8.6.1. Let Vα be a Whitney stratification of V and let f : V → C
be the restriction to V of a holomorphic function f : Cm → C; assume for
simplicity 0 = f (x). One says that f : (V, x) → (C, 0) has a stratified Morse
critical point at x ∈ V if the dimension of the stratum Vα that contains x is
≥ 1, the restriction of f to Vα has a Morse singularity at x and f is general
with respect to all other strata containing x in its closure, i.e., Ker df(x) is
transverse in Cm to every limit of tangent spaces Txi (Vβ ), for every stratum
Vβ such that Vα ⊂ V β and every sequence xi ∈ Vβ converging to x.
We recall that every map-germ f on (V, 0) can be morsified, i.e., approx-
imated by Morse singularities. This is proved in [104] for f with an isolated
singularity.
8.6 The Euler Obstruction via Morse Theory 141
The theorem below is contained in [150]. Notice that 8.5.1 is included here.
Remark: In [151] there are several formulae relating the invariant Euf,V (0)
with other invariants of functions on singular varieties. In [77] this invariant
is related with the Bruce–Roberts Milnor number defined in [38].
Chapter 9
Indices for 1-Forms
In this section we study some basic facts about the geometry of 1-forms and
the interplay between real and complex valued 1-forms on (almost) complex
manifolds, which plays an important role in the sequel. The material here is
all contained in the literature; we include it for completeness and to set up
thus
ω(v) = Re (ω)(v) − i Re (ω)(iv).
In other words the form ω is determined by its real part and one has a 1-to-1
correspondence between real and complex forms, assigning to each complex
form its real part, and conversely, to a real 1-form η corresponds the complex
form ω defined by:
ω(v) = η(v) − iη(iv).
This statement (observed in [50],[73]) refines the obvious fact that a com-
plex hyperplane P in Cm , say defined by a linear form H, is the intersection
:= {Re H = 0} and i H.
of the real hyperplanes H This justifies the following
definition:
Recall that the Euler class of an oriented vector bundle is the primary
obstruction to constructing a nonzero section [153]. In the case of the bundle
TR∗ M , this class equals the Euler class of the underlying real tangent bundle
TR M , since they are isomorphic. Thus, if M is compact then its Euler class
evaluated on the orientation cycle of M gives the Euler–Poincaré character-
istic χ(M ). We can say this in different words: let η be a real 1-form on M
with isolated (hence finitely many) singularities x1 , · · · , xr . At each xi this
η/η
1-form defines a map, Sε −→ S2m−1 , from a small (2m − 1)-sphere Sε in
M around xi into the unit sphere in the fiber (TR∗ M )x . If we equip M and
TR∗ M with the orientations induced by the almost complex structure on M ,
the degree of this map is the Poincaré–Hopf local index of η at xi , that we
may denote by IndPH (η, xi ). Then the total index of η in M is by definition
the sum of its local indices at the xi and it equals χ(M ). Its Poincaré dual
class in H 2m (M ) is the Euler class of TR∗ M ∼
= TR M .
More generally, if M is a compact C ∞ manifold of real dimension 2m with
nonempty boundary ∂M and a complex structure in its tangent bundle, one
can speak of real and complex valued 1-forms as above. Elementary obstruc-
tion theory (see [153]) implies that one can always find real and complex
1-forms on M with isolated singularities, all contained in the interior of M .
In fact, if a real 1-form η is defined in a neighborhood of ∂M in M and it
is nonsingular there, then we can always extend it to the interior of M with
finitely many singularities, and its total index in M does not depend on the
choice of the extension.
cm (T ∗ M ) = (−1)m cm (T M ).
This corresponds to the fact that at each isolated singularity xi of ω one has
two local indices: one of them is the index of its real part defined as above,
ω/ω
IndPH (Re ω, xi ); the other is the degree of the map Sε −→ S2m−1 , that we
denote by IndPH (ω, xi ). These two indices are related by the equality:
and the index on the right corresponds to the local Poincaré–Hopf index of
the vector field defined by duality near xi .
For example, the form ω = zi dzi
in Cm has index 1 at 0, while its real part (xi dxi − yi dyi ) has index (−1)m .
If we take M as above, compact and with possibly nonempty boundary,
and ω is a complex 1-form with isolated singularities in the interior of M and
radial on the boundary, then (by the previous considerations) the total index
of ω in M is (−1)m χ(M ). We summarize some of the previous discussion in
the following theorem ([49, 50]):
T ∗ Ux ∼
= π ∗ T ∗ U α ⊕ p ∗ T ∗ Dα .
V ); Z), and MacPherson defined the local Euler obstruction EuV (0) of V at
0 to be the integer obtained by evaluating this class on the orientation cycle
[ν −1 (Bε ∩ V ), ν −1 (Sε ∩ V )].
More generally, given a section η of TR∗ B|A , A ⊂ V , there is a canonical
way of constructing a section η̃ of TR∗ |Ã , Ã = ν −1 A, which is described in the
following. The same construction works for complex forms. First, taking the
pull-back ν ∗ η, we get a section of ν ∗ TR∗ B|V . Then η̃ is obtained by projecting
ν ∗ η to a section of T̃R∗ by the canonical bundle homomorphism
Thus the value of η̃ at a point (x, P ) is simply the restriction of the linear
map η(x) : (TR B)x → R to P . We call η̃ the canonical lifting of η.
By the Whitney condition (a), if a ∈ Vα is the limit point of the sequence
{xi } ∈ Vreg such that P = lim(T Vreg )xi and if the kernel of η is transverse
to Vα , then the linear form η will be nonvanishing on P . Thus, if η has an
isolated singularity at the point 0 ∈ V (in the stratified sense), then we have
a never-zero section η of the dual Nash bundle TR∗ over ν −1 (Sε ∩ V ) ⊂ V .
Let o(η) ∈ H 2n (ν −1 (Bε ∩ V ), ν −1 (Sε ∩ V ); Z) be the cohomology class of the
obstruction cycle to extend this to a section of TR∗ over ν −1 (Bε ∩ V ). Then
define (c.f. [32, 50]):
Definition 9.3.1. The local Euler obstruction of the real differential form
η at an isolated singularity is the integer EuV (η, 0) obtained by evaluating
the obstruction cohomology class o(η) on the orientation cycle [ν −1 (Bε ∩
V ), ν −1 (Sε ∩ V )].
T ∗ B|A with an isolated singularity, one can define the local Euler obstruction
EuV (ω, 0). Notice that it is equal to that of its real part up to sign:
Theorem 9.3.2. Let Vα ⊂ V be the stratum containing 0, EuV (0) the local
Euler obstruction of V at 0 and ω a (real or complex) 1-form on Vα with
an isolated singularity at 0. Then the local Euler obstruction of the radial
extension ω of ω and the Schwartz index of ω at 0 are related by the following
proportionality formula:
The Theorem can be proved by one of the two ways we used for proving
Theorem 3.6.1 or Theorem 8.1.2 (for details see [36]).
Let us fix a radial vector field vrad on (V, 0), e.g., the gradient on the
smooth part of V of the real valued function z with respect to a Riemannian
metric.
Definition 9.4.1. A real (or complex) 1-form on V is radial at 0 if, near the
origin, its value on the radial vector field vrad has positive real part at each
point in a punctured neighborhood of the origin 0 in V . We denote such a
form by ωrad .
The space of such 1-forms is connected.
Let ω1 and ω2 be 1-forms on (V, 0) with isolated singularities at the origin.
Choose ε > ε > 0 sufficiently small, let Kε = V ∩Sε and Kε = V ∩Sε be the
corresponding links, and let Z be the cylinder V ∩ [Bε \ Int(Bε )], where Bρ
is the ball of radius ρ around the origin 0 in Cm , Sρ is its boundary. Let ω be
a 1-form on the cylinder Z which coincides with ω1 in a neighborhood of Kε
and with ω2 in a neighborhood of Kε and which has isolated singular points
q1 , . . . , qs inside Z. The sum d(ω1 , ω2 ) of the (usual) local indices Ind(
ω , qi )
of the form ω at these points depends only on the forms ω1 and ω2 and will
be called the difference of these forms. One has d(ω1 , ω2 ) = −d(ω2 , ω1 ).
Remark 9.4.1. Notice this definition is similar to that of the Schwartz index
of vector fields given in the first section of Chap. 2. Also notice that the index
of a radial 1-form ωrad is equal to (−1)n . The sign is chosen so that this index
coincides with the usual one if V is smooth at 0.
Remark 9.4.2. We know from Sect. 1 in this chapter that there is a one-to-
one correspondence between complex 1-forms on a complex analytic manifold
V \ {0} and real 1-forms on it. The radial index of a complex 1-form can be
expressed through the corresponding index of its real part, defined in [50,51],
and viceversa. As before, the radial index Indrad (ω, 0; V ) of a complex 1-form
ω equals (−1)n -times the radial index of its real part.
where the sum on the right hand side is over all those points x in a small
punctured neighborhood of the origin 0 in V where the form ω vanishes (this
follows from the fact that d(ω1 , ω3 ) = d(ω1 , ω2 ) + d(ω2 , ω3 )). This stability
property of the index will be used in the last section.
Remark 9.4.4. In [52] Ebeling and Gusein-Zade define the radial index in a
more general setting, analogous to the way we defined this index for vector
fields in 2.4.2, and the theorem below holds in that more general setting.
independently of the 1-form. This is actually a special case of the last state-
ment in Theorem 9.1.1, taking the boundary to be empty. We also know
that in the case of vector fields, the work of M.-H. Schwartz shows that if
V is now a compact complex analytic singular variety and v is a stratified
vector field on V obtained by radial extension, then its total index equals
the Euler–Poincaré characteristic χ(V ). As explained in Chap. 2, this result
extends easily to arbitrary vector fields on V , provided they are stratified
and with isolated singularities, using the radial index.
These results extend naturally to 1-forms on singular varieties, as observed
by W. Ebeling and S. Gusein-Zade in [50, 51]. One gets:
We look first at the case studied by Ebeling and Gusein-Zade, i.e., when the
variety V is an isolated complete intersection germ; then we envisage the case
when V has nonisolated singularities, following [36].
154 9 Indices for 1-Forms
from the link K of V into the Stiefel manifold Wk+1 (n+k), where (f1 , . . . , fk )
are functions that define the ICIS germ (V, 0) and v is assumed to be nonzero
away from 0. As we know, this index equals the Poincaré–Hopf index of an
extension of v to a Milnor fiber F. The analogous index was defined in [50,51]
for 1-forms on V .
Let ω be a complex-valued 1-form on V with an isolated singularity at 0.
Then its GSV index equals the degree of the map:
∗
ψω = (ω, df1 , . . . , dfk ) : K → Wk+1 (n + k),
∗
where Wk+1 (n + k) denotes the bundle associated to the cotangent bun-
∗ n+k
dle T (C )|V with fiber the corresponding Stiefel manifold of complex
orthonormal (k + 1)-frames in the dual of Cn+k . As noticed in [50, 51], this
index equals the Poincaré–Hopf index of the 1-form on a Milnor fiber of f ,
i.e., it equals the number of zeroes, counted with multiplicities, of any exten-
sion of ω to a Milnor fiber Vt = f −1 (t) ∩ Bε of (V, 0). (The proof is similar to
that for vector fields given in Chap. 3.)
A remarkable difference of this index with the analogous one for vector
fields was observed in [50, 51]: if the differential 1-form is holomorphic, then
its index can be regarded as an intersection number of complex manifolds,
while for vector fields, the definition of the GSV index involves the conjugate
gradient vector fields, which are anti-holomorphic. Thence, in the case of
holomorphic 1-forms we can use powerful techniques of algebraic geometry
to compute its index. More precisely, assume the 1-form ω is holomorphic, and
let I be the ideal in OCn+k,0 generated by f1 , . . . , fk and the (k + 1) × (k + 1)-
minors of the matrix: ⎛ ∂f1 ⎞
∂x1 · · · ∂xn+k
∂f1
⎜ . . ⎟
⎜ . ⎟
⎜ . · · · .. ⎟ .
⎜ ∂fk ∂fk ⎟
⎝ ∂x1 · · · ∂xn+k ⎠
A1 · · · An+k
Then one has the following theorem of W. Ebeling and S. Gusein-Zade (see
[50, 51]):
Theorem 9.5.1.
This formula extends the one obtained by Lê D.T. and G.-M. Greuel for
the case when ω is the differential of a function ([74, 101]); in that case the
formula gives the Milnor number of the function and is known as the Lê-
Greuel formula for the Milnor number.
Remark 9.5.1. The above index can be regarded in the more general setting of
residues of Chern classes defined by a finite number of holomorphic sections.
See [160], where various expressions of the residues as in Sect. 1.6.6 are given.
The formula (9.5.1) is a particular case of the algebraic expression there.
It is clear that the GSV index satisfies the same law of conservation of
number satisfied by the radial index: if ω is a 1-form on V close to ω, then:
IndGSV (ω, 0; V ) = IndGSV (ω , 0; V ) + Indrad (ω , x; V ),
where the sum on the right hand side is over all those points x in a small
punctured neighborhood of the origin 0 in V where the form ω vanishes.
This implies:
Proposition 9.5.1. Let μ(V, 0) be the Milnor number of the isolated com-
plete intersection singularity (V, 0). For any 1-form ω on (V, 0) with an
isolated singularity at the origin 0 one has
In other words this index measures the number of points (counted with
signs) in which a generic perturbation of ω is tangent to F. In fact the in-
i
clusion F → M pulls the form ω to a section of the (real or complex, as the
case may be) cotangent bundle of F, which is never-zero near the boundary
∂F since ω has an isolated singularity at 0 and, by hypothesis, the map f
satisfies the Thom af -condition. One gets the following result, which is due
to W. Ebeling and S. Gusein-Zade [50] when V has an isolated singularity:
where e(F; ω) ∈ H 2n (F, ∂F) is the Euler class of the real cotangent bundle
TR∗ F relative to the section defined by ω on the boundary, and [F] is the
orientation cycle of the pair (F, ∂F). If ω is a complex form, then one has:
Notice this is analogous to the construction done in Sect. 1.3.2. In this case
one can, alternatively, express the index as the relative Chern class:
Remark 9.5.2. We notice that Theorems 9.3.2 and 9.5.6 can also be proved
using the stability of the index under perturbations, just as we did for vec-
tor fields. More precisely, one can easily show that the Euler obstruction
EuV (ω, x) and the GSV index are stable when we perturb the 1-form (or the
vector field) in the stratum and then extend it radially; then the sum of the
indices at the singularities of the new 1-form (vector field) give the corre-
sponding index for the original singularity. This implies the proportionality
of the indices.
158 9 Indices for 1-Forms
This and the following sections are taken from [57]. Here we introduce the
homological index of a 1-form on a complex analytic variety with an isolated
singular point. This is analogous to, and inspired by, the homological index
for vector fields defined in [68] and discussed in Chap. 7. As in the case of
vector fields, when the ambient space is an ICIS, this index coincides with
the previously defined GSV index of 9.5.1.
Let (V, 0) ⊂ (Cm , 0) be an arbitrary germ of an analytic variety of pure
dimension n with an isolated singular point at the origin (not necessarily a
complete intersection). Given a holomorphic form ω on (V, 0) with an isolated
•
singularity, we consider the complex (ΩV,0 , ∧ω):
0 −→ OV,0 −→ ΩV,0
1
−→ · · · −→ ΩV,0
n
−→ 0,
i
where ΩV,0 are the modules of germs of Kähler differential forms on (V, 0) as
in 7.1.1, and the arrows are given by the exterior product by the form ω.
This complex is the dual of the Koszul complex considered in Chap. 7, and
it was used by G.-M. Greuel in [74] for complete intersections. The sheaves
•
i
ΩV,0 are coherent sheaves and the homology groups of the complex (ΩV,0 , ∧ω)
are concentrated at the origin and therefore are finite dimensional.
n
•
Indhom (ω, 0; V ) = (−1)n−i hi (ΩV,0 , ∧ω), (9.6.1)
i=0
•
where hi (ΩV,0 , ∧ω) is the dimension of the corresponding homology group as
a vector space over C.
where the sum on the right hand side is over all those points x in a small
punctured neighborhood of the origin 0 in V where the form ω vanishes.
9.7 On the Milnor Number of an Isolated Singularity 159
class map (see [41]). In the article [123] of D. Mond and D. Van Straten
there is considered a Milnor number of a function f on a curve singular-
ity introduced by V. Goryunov. One can see that this Milnor number can
be defined for a 1-form ω with an isolated singularity on (C, 0) as well (as
dimC (ωC,0 /ω ∧ OC,0 )) and is equal to Indhom ω + λ − τ .
between the homological and the radial indices does not depend on the
1-form ω.
In other words this theorem says that the radial index (which is defined
topologically) equals the difference between the Euler characteristics of the
usual de Rham complex and the complex given by multiplication by the
1-form ω. This might be a special case of a general theorem for singular
varieties in the spirit of the results of C. Simpson [152] and others for complex
manifolds.
The idea of the proof is to consider the normalization π : (C̄, 0̄) → (C, 0)
of the curve and the commutative diagrams:
9.8 Indices for Collections of 1-Forms 161
∧ω
0 → O⏐
C,0 −−−→ Ω⏐
C,0 −−−
1
→ ΩC,0
1
/ω⏐∧ OC,0 → 0
⏐ ∗ ⏐ ∗ ⏐
⏐ π0 ⏐π1 ⏐
∧ω̄
0 → OC̄,0̄ −−−→ ΩC̄,
1
0̄
−−−→ ΩC̄,
1
0̄
/ω̄ ∧ OC̄,0̄ → 0
d
C,0 −−−→
0 → m⏐ C,0 → ΩC,0 /dOC,0 → 0
1
Ω⏐ 1
⏐ ∗ ⏐ ∗
⏐π 0 ⏐ π1 ↓
d
0 → mC̄,0̄ −−−→ ΩC̄,
1
0̄
→ 0
where mC,0 is the maximal ideal in the ring OC,0 , mC̄,0̄ is the ideal of germs
of functions on the normalization (C̄, 0̄), equal to zero at all the points in 0̄.
Then the snake Lemma yields to 9.7.1 (see [57, 4.3] for the complete proof).
We also know from the previous chapters that one has similar statements
for vector fields on compact complex analytic varieties, the precise statement
one gets depending on the concept of index one is using. In fact, as we shall
see in Chaps. 10–13 of this monograph, these are related to various concepts
of “Chern classes” one has for singular varieties, which coincide with the usual
Chern classes in the case of manifolds. For the radial index one gets χ(V ),
this is the 0-degree Schwartz–MacPherson class of V . For the GSV-index one
gets the 0-degree Fulton–Johnson class of V , which (with some restrictions)
equals the Euler–Poincaré characteristic of a smoothing V as in the proof of
162 9 Indices for 1-Forms
Theorem 3.2.2. For the local Euler obstruction one gets the top Chern class
of the Nash bundle over the Nash blow up of V .
Now, in the case of manifolds the number cm (M )[M ] is one of the Chern
numbers the manifold has, but there are several others. One has a Chern
number
cj1 (M ) cj2 (M ) · · · cjr (M ) [M ],
whenever j1 , ..., jr are positive integers which add up to m. In the next chap-
ters of this monograph we shall explore various ways of generalizing the Chern
classes of complex manifolds to the case of singular varieties, and as we shall
explain, these are related in one or another way to studying indices of ap-
propriate vector fields or frames, as described in Chap. 1. These yield to
homology (or cohomology) classes which represent various generalizations
of Chern classes for singular varieties (of course there can be other means
to constructing Chern classes for singular varieties, for instance using the
MacPherson functor [117]).
Yet, there is another question that arises naturally in the context of this
book: the cj1 (M ) are cohomology classes in H 2ji (M ; Z), but the evaluation
cj1 (M )cj2 (M ) · · · cjr (M )[M ] is an actual number, in fact an integer. Is there a
way of defining an index associated to this number in a similar way as the local
Poincaré–Hopf index of a vector field (or 1-form) is associated to the Chern
number cm (M )[M ]? Moreover, what can we say about this question when the
ambient space is now a singular variety? what information these invariants
give about singular varieties? These and other questions are addressed by
W. Ebeling and S. Gusein-Zade in a series of articles (see [53–56]).
Before looking at this matter, let us envisage some related facts about the
Chern numbers of manifolds.
We recall from Chap. 1 that given a complex manifold M of dimension m,
its Chern class cr (M ) ∈ H 2r (M ) is the primary obstruction to constructing
an (m − r + 1)-frame in M . In other words, let Wm−r+1 (m) be the Stiefel
manifold of complex orthonormal (m − r + 1)-frames in Cm . This manifold
is diffeomorphic to U (m)/U (r − 1) and therefore it is (2r − 2)-connected and
its first nonzero homology and homotopy groups are H 2r−1 (Wm−r+1 (m)) ∼ =
π2r−1 (Wm−r+1 (m)) ∼ = Z (see [153]).
Now let Wm−r+1 (m)(T M ) denote the fibre bundle over M whose fibre
at each point x is the Stiefel manifold Wm−r+1 (m) of complex orthonormal
(m − r + 1)-frames in Tx M ∼ = Cm . Let us try to construct a section of this
bundle via the usual stepwise process. We triangulate M in some (any) way
and construct a section of Wm−r+1 (m)(T M ) step by step, starting from the
0-skeleton, then the 1-skeleton and so on, as far as we can. The fact that
the fiber is (2r − 2)-connected tells us that we can construct such a section
up to the (2r − 1)-skeleton of the triangulation. The first possibly nonzero
obstruction arises when we try to extend the section over the 2r-skeleton. We
thus get an element in π2r−1 (Wm−r+1 (m)) ∼ = Z associated to each 2r-cell (or
simplex). This defines a cochain of dimension 2r, which is actually a cocycle
9.8 Indices for Collections of 1-Forms 163
are the points that count for the Chern number cr cm−r [M ]. And the way
each singular point contributes towards this Chern number in the example
above is ±1 because of the transversality assumptions we made. In general
this is an integer that can be regarded as a local index associated to the
corresponding collection of vector fields at each singular point.
Now suppose we are given integers r1 , ..., rs such that r1 + ... + rs = m.
One has a Chern number cr1 (M ) · · · crs (M ) [M ], and the previous discussion
extends to this setting, to say that this number is the intersection number of
the cycles representing the Poincaré duals of the corresponding Chern classes.
Each Chern class crj corresponds, by duality, to the set of points where a
certain family of vector fields becomes linearly dependent, the singularities
of the corresponding frame v m−rj +1 . A point x ∈ M is nonsingular for the
collection of vector fields
164 9 Indices for 1-Forms
There are two main situations we consider, following [53–56]. One of this
leads to a generalization of the GSV-index, the other to a generalization of
the Euler defect.
Assume now that (V, 0) is the germ of a reduced complex analytic space in
Cm of pure dimension n, and equip V with a Whitney stratification so that
{0} is a stratum. If ω is a 1-form on Cm with an isolated singularity at 0, then
one has its local Euler obstruction defined in 9.3.1 above. We recall briefly
its definition. Let V → V be the Nash blow up of V , and T∗ → V the dual of
ν π
its Nash bundle. Then the 1-form ω lifts canonically to a section ω of T∗ over
166 9 Indices for 1-Forms
We know already from Chap. 2 how to define the Schwartz index of a vector
field constructed by radial extension, and that notion was generalized to give
an index for vector fields in general, using the difference between the given
vector field and a radial one (2.4.2). In this section we extend these concepts
to frames, using the difference cocycle (see [153]), in order to define the local
Schwartz index for arbitrary stratified frames.
The idea for constructing frames by radial extension is similar to that for
constructing vector fields by radial extension, that we described in Chap. 2
(see [142]). We consider as before, a compact, complex analytic n-dimensional
variety V embedded in a complex m-manifold M , endowed with a Whitney
stratification {Vα } adapted to V . We use a cellular decomposition (D), dual
to a triangulation of M compatible with the stratification. The cells σ of (D)
are transverse to the strata Vα . In general, elements σ ∩ Vα of (D) are not
cells, but that is the case for the smallest dimensional stratum Vα meeting σ.
The concept of stratified vector fields, introduced in Chap. 2, extends in
the obvious way to frames:
Definition 10.1.1. Let A be a subspace of M . A stratified r-frame on A is
an r-field v (r) = {v1 , ..., vr } consisting of stratified vector fields v1 , . . . , vr , lin-
early independent everywhere. By a singularity of an r-frame we mean a point
z ∈ A where the r vectors v1 (z), . . . , vr (z) fail to be linearly independent.
Let σ be a (D)-cell of dimension 2(m − r + 1), dual of a simplex in the
(complex) d-dimensional stratum Vα . Then σα = σ ∩ Vα is a cell of (real)
dimension 2d − 2r + 2. Suppose we have a stratified r-frame v (r) defined
on the boundary of σα . We thus have an associated map v (r) : ∂σα →
Wr,d , which determines an element in π2d−2r+1 (Wr,d )
Z. This defines an
index Ind(v (r) , σα ) ∈ Z, as in Chap. 1, that by abuse of notation we call the
Poincaré–Hopf index of the frame v (r) on σα . The frame can of course be
extended by a homothecy to all of σα minus its barycenter aσ , Notice that
if we write the frame v (r) as (v (r−1) , vr ), where v (r−1) is the (r − 1) frame
determined by the first r − 1 vector fields in v (r) , then v (r−1) extends without
singularities to the interior of σα , by dimensional reasons. It spans a bundle
Sp{v (r−1) }. Then the index Ind(v (r) , σα ) equals the degree of the map from
∂σα into the unit sphere of the fiber over aσ of orthogonal complement of the
bundle Sp{v (r−1) }, defined by the last vector field vr (normalized).
The M.-H. Schwartz’s radial extension technique can be used to extend
each of the components of v (r) in a neighborhood of σα in M (see [28]). In
this way we obtain a stratified r-frame on a punctured neighborhood of aσ
in the 2(m − r + 1)-cell σ of (D). The boundary of σ is a (2m − 2r + 1)-
sphere. This defines a map ∂σ → Wr,m which represents an element in
π2m−2r+1 (Wr,m )
Z and therefore defines an index in Z. Since by construc-
tion the frame is radial in all directions normal to the stratum Vα , it follows
that this index coincides with the previously defined index Ind(v (r) , σα ). One
has:
Theorem 10.1.1. Let Vα be a (complex) d-dimensional stratum in V and
suppose σα = σ ∩ Vα is a (2d − 2r + 2)-cell. Let v (r) be an r-frame defined on
the boundary of σα Let us write v (r) = (v (r−1) , vr ) where v (r−1) is the (r − 1)
frame determined by the first r − 1 components (vector fields) in v (r) . Then:
10.1 The Local Schwartz Index of a Frame 169
∂ ∂ ∂
v (3) = ( , , z k ), k ≥ 1.
∂x ∂y ∂z
Using the second definition above we see that the index of this frame equals
the Poincaré–Hopf index at 0 ∈ C of the holomorphic vector field z k ∂z ∂
, so it
2
has index k. Notice that we may consider only a 2-frame on e , for example
∂
v (2) = ( ∂x ∂
, z k ∂z ); this has a singularity at {(0, 0, 0)}, but in this case one
can get rid of this singularity by deforming the frame by an appropriate
homotopy, since the 2-frame on the boundary extends to the interior if and
only if the corresponding map into W2,3 is nulhomotopic, but this Stiefel
manifold is diffeomorphic to S3 , so it is simply connected. In the previous
case the 3-frames on ∂e2 = S1 correspond to the elements in the fundamental
group of the unitary group U (3)
W3,3 , π1 (U (3))
Z.
∂ ∂ ∂ ∂
v (2) = (x + ,x + zk ), k ≥ 1,
∂x ∂y ∂x ∂z
Now we wish to distinguish the normally radial stratified frames with the
“simplest” possible behavior in the stratum Vα of aσ . Write such a frame v (r)
in the form (v (r−1) , vr ) as before; we assume the frame v (r−1) is extended to
the cell σ with no singularity, so the only singularity of v (r) in σ is that of
the vector field vr at the barycenter aσ . Let Q ⊂ T M |σ be the orthogonal
complement (for some Riemannian metric) of the bundle Sp{v (r−1) }; the
Schwartz index of v (r) at aσ is the Poincaré–Hopf index of the section vr of Q.
frame ∂ ∂ ∂ ∂
x + ,x +z ,
∂x ∂y ∂x ∂z
is radial in the stratum e and it is also normally radial, so it is radial in the
ambient space.
We now define the local Schwartz index for arbitrary (stratified) frames;
this is similar to 2.4.2. Let v (r) be an r-frame defined on the boundary of
a (D)-cell σ of dimension 2m − 2r + 2, whose barycenter is a point aσ ∈
Vα ⊂ V . We extend v (r) to a stratified frame on all of σ \ {aσ }. Recall that,
by construction, the cell σ meets transversally all the Whitney strata Vβ
(r)
containing Vα in their closure. Let vrad be a stratified radial frame around
(r)
aσ . We define the difference between v (r) and vrad at aσ as follows. Consider
sufficiently small spheres Sε , Sε in M , ε > ε > 0, centered at aσ , and
10.2 Proportionality Theorem 171
(r)
consider the frame v (r) on Sε ∩ σ ∩ V and vrad on Sε ∩ σ ∩ V . We use again
the Schwartz’s technique of radial extension to get a stratified r-frame w(r)
on the intersection of σ with the cylinder
◦
X = [(V ∩ Be ) \ (V ∩ Be )]
(r)
d(v (r) , vrad ) = IndPH (w(r) , X ∩ σ),
where the sum on the right runs over the singular points of w(r) in X and each
singularity is being counted with the local index of w(r) in the corresponding
stratum. As in the work of M.-H. Schwartz, we can check that this integer
does not depend on the choice of w(r) .
Definition 10.1.5. The Schwartz (or radial) index of the stratified r-field
v (r) at aσ ∈ V is:
(r)
IndSch (v (r) , aσ ; V ) = 1 + d(v (r) , vrad ) = 1 + Ind(w(r) , X ∩ σ),
where the sum on the right is taken over the singularities of w(r) in all strata
in X ∩ σ.
The Proportionality Theorem, due to [33], is the key point in the proof of the
equality of Schwartz and MacPherson classes via the Alexander isomorphism.
Let ν : V → V be the Nash modification of V . Let σ be a cell of dimension
2(m − r + 1) and v (r) a stratified r-field on σ ∩ V with an isolated singularity
at the barycenter aσ of σ. Since v (r) is nonsingular on ∂σ ∩ V , it can be lifted
to an r-frame v(r) of T over ν −1 (∂σ ∩ V ), as in the case of vector fields. One
obtains, over ν −1 (∂σ ∩ V ), a section of the bundle Tr associated to T and
whose fiber in a point x̃ is the set of r-frames in Tx̃ . Let o( v (r) ) denote the
2(n−r+1) −1 −1
class in H (ν (σ ∩ V ), ν (∂σ ∩ V )) of the obstruction cocycle to
extending v further to a section of Tr |ν −1 (σ) .
(r)
Definition 10.2.1. The local Euler obstruction EuV (v (r) , aσ ) of the strat-
ified r-field v (r) at the isolated singularity aσ is the integer obtained by
v (r) ) on the orientation cycle [ν −1 (σ ∩ V ), ν −1 (∂σ ∩ V )].
evaluating o(
172 10 The Schwartz Classes
A possible proof is to use the stability of the Euler obstruction EuV (v (r) , a)
under appropriate perturbations and to prove the Theorem in the same way
as Theorem 3.6.1. We give a proof adapted from the one of [33] but which is
simpler (see [35]).
T M |σ = E ⊕ Q. (10.2.2)
T M |Ω = E ⊕ Q. (10.2.3)
v0 , T)[ν −1 (Ω ∩ V ), ν −1 (S ∩ V )].
Eu(v0 , V ; p) = o(
This will finish the proof of Theorem 10.2.1, since Theorem 8.1.2 implies
ν ∗ T M |σ = ν ∗ E ⊕ ν ∗ Q.
where ! denotes the cup product. We have o(ν ∗ π1∗ vD , ν ∗ E) = ν ∗ π1∗ o(vD , T D).
Since Ω ∩ V = D × (σ ∩ V ) and o(vD , T D) is a generator of H 2r−2 (D, ∂D), we
get
where " denotes the cap product. Since the restriction of o(π&
∗ −1
2 vr , P ) to ν σ
is equal to o(vr , P ), we obtain
Remark 10.2.1. Notice that both proofs above are easily adapted to prove
the equivalent theorem for coframes on V , i.e., for frames of 1-forms on V .
This gives relations between the dual Schwartz classes of V (which can be
represented by Chern classes of the cotangent bundle T ∗ (U ), U being now
a regular neighborhood of V in M , relative to a coframe on U \ V obtained
by radial extension) and the corresponding Chern classes of the dual Nash
bundle T∗ . This is related to the recent work of Ebeling and Gusein-Zade
about indices of “collections” of 1-forms [55].
10.3 The Schwartz Classes 175
Corollary 10.3.1. With the above hypothesis and notation, if there exists
some stratified r − f rame which is nonsingular on (D)(2m−2r+2) , then the
corresponding Schwartz class of V vanishes.
Since H 2q (M, M \ V )
H2m−2q (V ) by Alexander duality, these classes
can be thought of as living in the homology of V . We denote by ci (V ),
i = m − q = n − p, the image of cq (V ) in H2i (V ) and call it the homology
Schwartz class.
The basic reference for this section is [25] (see also [28] and [31]). Here we
adapt to the singular case the discussion started in Sect. 1.3 concerning the
Alexander homomorphism and other related topics.
Let V be a pure n-dimensional subvariety in an m-dimensional complex
manifold M and let (K), (K ) and (D) be as before, (K) is a triangula-
tion of M adapted to V , (K ) its first barycentric subdivision and (D) the
corresponding dual cell decomposition.
First, if V is compact, we define a homomorphism
(K)
) (V ) −→ Ci c, d(s) ∩ V
2n−i
P : C(K (V ) by P (c) = (10.4.1)
σ
PV : H 2n−i (V ) −→ Hi (V ),
(K)
) (V, V \ S) −→ Ci
2n−i
A : C(K (S)
which is called the Gysin homomorphism. From the definition, we see that,
if M is compact, the following diagram is commutative:
GV,M
H 2n−i (V ) −−−−→ H 2m−i (M )
⏐ ⏐
⏐P ⏐
V PM (10.4.4)
i
Hi (V ) −−−∗−→ Hi (M ).
) (V ) −→ C(D) (M, M \ V ).
2n−i 2m−i
T : C(K (10.4.5)
which is called the Thom homomorphism. From the definition, we see that,
if V is compact, the following diagram is commutative:
TV,M
H 2n−i (V ) −−−−→ H 2m−i (M, M \ V )
⏐ ⏐
⏐P ⏐ (10.4.7)
V AM,V
=
Hi (V ) −−−−→ Hi (V ).
178 10 The Schwartz Classes
which is also called the Thom homomorphism. From the definition, we see
that the following diagram is commutative:
TS,V,M
H 2n−i (V, V \ S) −
−−−−
→ H 2m−i (M, M \ S)
⏐ ⏐
⏐A ⏐ (10.4.10)
V,S AM,S
=
Hi (S) −−−−→ Hi (S).
It follows from the above discussion that there exist stratified r-fields on
(r) (r)
(D)(2q) ∩ U whose singularities are all located on S. Let v1 and v2 be two
such r-fields and let us consider a tube T in U around S. There is a well
(r) (r)
defined secondary characteristic class d(v1 , v2 ) ∈ H 2p−1 (∂T ) called the
difference and defined as in Sect. 1.3.2. Let δ : H 2p−1 (∂T ) −→ H 2p (T , ∂T )
be the connecting homomorphism and let AV : H 2p (T , ∂T ) −→ Hr−1 (S) be
the Alexander homomorphism. We set
where the sum is taken over the connected components of Sing(V ). In partic-
(r)
ular, for a radial r-frame vrad , we have:
(iλ )∗ cr−1 (Sλ ) + ι∗ cr−1 (V ∗ ; vrad ).
(r)
cr−1 (V ) =
Sλ ⊂Sing(V )
Remark 10.5.1. In other words this theorem is telling us that to define the
Schwartz class cq (V ) ∈ H 2q (M, M \ V ; Z) we may consider any stratified
r-frame v (r) , q = m − r + 1, on (D)(2q) ∩ U, then cq (V ) is the obstruction
to extending it to a stratified r-frame on V ∩ (D)(2q) . The contributions for
cq (V ) are splitted in two parts. On one hand we have the contribution of
the regular part ι∗ cr−1 (V ∗ ; v (r) ); this is the usual Chern class of V minus
an open regular neighborhood of Sing(V ) relative to the choice of frame v (r)
on its boundary (in the appropriate skeleton). On the other hand we have
the individual contributions of each connected component of the singular set
(iλ )∗ Sch(v (r) , Sλ ). Each of these depends on the choice of frame, but their
total sum is cq (V ) independent of the frame, a result in the spirit of the
Poincaré–Hopf Theorem for vector fields.
Remark 10.5.2. Again, just as in 2.4.4, one may consider a stratified r-frame
v (r) on the intersection of (D)(2q) with a neighborhood U ⊂ M of a con-
nected component S ⊂ Sing(V ) whose singularities are all in S. We have
defined above a localization Sch(v (r) , S) of the Schwartz class cr−1 at S. But
in the previous section we defined a local index for v (r) at each isolated sin-
gularity. Taking into account only the singularities of v (r) in S we get another
, (r) , S). The proof of 2.4.4 can be easily
localization of cr−1 at S, say Sch(v
adapted to this case, thus showing that both localizations of the Schwartz
class coincide.
for a relative cycle γ ∈ C2q (T , ∂ T ) [108, 109], where Ck (A) denotes
(D) (D)
(r)
, U
H 2q (U \ S) is the q-th Schwartz class of S. We may assume that v (r) is
rad
given on a neighborhood W of (U \ S) ∩ D(2q) . We denote by ∇ a connection
for T M on U and by ∇ 0 an v (r) -trivial connection for T M on W . From the
0
definitions, we have the following.
(r)
in U . For each i = 1, 2, let ∇i be an vi -trivial connection for T V0 on W .
We refer [31, Lemma 3.4] for the proof of the following
(r) (r)
Lemma 10.5.1. The difference δd(v1 , v2 ) is in H 2p (U, U \S) whose image
by the Thom–Gysin homomorphism τ : H 2p (U, U \ S) → H 2q (U , U
\ S) is
represented by the cocycle
γ → cp (∇1 , ∇2 ),
γ∩∂R
Let us discuss briefly the MacPherson classes (see [117], [28]). We use the local
Euler obstruction, defined in Chap. 8. This obstruction satisfies the following
property: there exists (unique) integers {nα } for which the equation
nα EuV α (x) = 1 (10.6.1)
is satisfied for all points x in V , where the sum runs over all strata Vα
containing x in their closure. This statement is obvious for points in the
regular stratum, for the points in Sing(V ) (10.6.1) can be easily proved by
induction.
Consider now (see Chap. 8) the Nash blow up V −→ V of V , the Nash
ν
bundle T −→ V , and the Chern classes of T, cj (T) ∈ H 2j (V ). The Poincaré
π
∩[V ]
βV : H 2j (V ) −→ H2n−2j (V ),
carries these into homology classes which can be pushed forward into the
homology of V via the homomorphism ν∗ induced by the projection.
The relation between Mather classes on one side and MacPherson classes
on the other side follows form MacPherson’s definition itself: his construc-
tion uses Mather classes, taking into account the values of the local Euler
obstruction along the strata. The precise relation between Mather classes
and Schwartz classes was determined in [33] (see 10.6.2 below), and this is
a key point for the identification of Schwartz and MacPherson classes. The
MacPherson classes live in H∗ (V ). It is proved in [33] that the Alexander ho-
momorphism H j (M, M \ V ) → H2m−j (V ) carries the Schwartz classes into
MacPherson’s. A key step in that proof is the Proportionality Theorem of
[33], that we discussed in Sect. 10.2.
Notice that the Schwartz indices of frames determine the elementary cocy-
cle in 10.3 that defines the corresponding Schwartz class. Thus the theorem
above establishes a deep connection between Schwartz classes and Mather
classes. This relation is made precise in the following theorem of [33, Th.
4.1] that we state without proof (for a complete proof, see [28]). Recall that
the Euler obstruction is constant on Whitney strata ([33, 10.2]), so we write
EuVα to denote the Euler obstruction at points in the stratum Vα .
Once we have this expression for a cycle representing the Mather classes
as weighted Schwartz classes, the identification of Schwartz and MacPherson
classes follows (see [28, 33]).
Since the definition of Schwartz classes and MacPherson classes is com-
pletely different, the fact that they coincide brings a deep richness into the
subject. Schwartz classes are geometrically defined and allow us to under-
stand what these classes measure in terms of obstruction theory, but they
are hard to work with. On the contrary MacPherson classes are in a way
more difficult to grasp, but they have useful factorial properties. This makes
184 10 The Schwartz Classes
them be powerful invariants and easy to work with. These important invari-
ants of singular varieties have been widely studied by many authors. We refer
to [28] for a complete account on the subject.
In the sequel we refer to these as Schwartz–MacPherson classes, or just SM-
classes for short, and we denote them cSM SM
∗ . The class ci lives in H2i (V ; Z),
i = 0, · · · , n.
Chapter 11
The Virtual Classes
The total Chern class of the virtual tangent bundle is defined in the usual
way:
c∗ (τV ) = c∗ (T M |V ) ! c∗ (N |V )−1 in H ∗ (V ) .
The pth Chern class of τV is the component of c∗ (τV ) in dimension 2p, for
p = 1, ..., n. That is, cp (τV ) is the coefficient of tp in the expansion of
m
k
(1 + ti ci (T M |V )) (1 + tj cj (N |V ))−1 .
i=1 j=1
r∗ ◦ ι∗ : H∗ (Vt ) −→ H∗ (V ),
In this section we define an index for stratified frames generalizing the GSV
index defined in Chap. 3 for vector fields on hypersurface germs with noniso-
lated singularities.
Let us denote by U an open neighborhood of the origin in Cn+1 , n > 0
and let us consider a holomorphic function
for every ε > 0 sufficiently small and δ = δ(ε) > 0 sufficiently small with
respect to ε, where Bε is a small ball around 0 ∈ Cn+1 and Dδ is a small ball
around 0 ∈ C.
We know already from Chap. 3 that given a stratified vector field v on V
with an isolated singularity at a point x ∈ V , we can lift v to a local Milnor
fiber F(x) of V at x and the Poincaré–Hopf index of this lifting to F(x) is well
defined and gives an invariant of v at x ∈ V that we called the GSV index.
We proved that if v is obtained by radial extension then its GSV index at
x is proportional to the Schwartz index of v at x, the proportionality factor
being the Euler characteristic of F(x) (Theorem 3.6.1). We are going to make
the analogous constructions for frames.
Let us recall the tube map π : F → V , from a local Milnor fiber into V ,
introduced in 3.5.3, that we use to lift vectors from V to F. We do it only
for hypersurfaces. Consider a small ball Bε around 0 ∈ U . Let ρ be a radial
vector field in a sufficiently small (with respect to ε) disk Dδ around 0 ∈ C,
whose solutions are arcs converging to 0. We can assume further that for each
t ∈ Dδ \ {0} the (Milnor) fiber Ft = f −1 (t) intersects the boundary sphere
Sε = ∂Bε transversely. Set T = f −1 (Dδ \ {0}). As we know,
f |T : T −→ Dδ \ {0}
is a locally trivial fiber bundle, and by [167] we can lift ρ to a rugose (hence
integrable) vector field ρ in T , whose solutions are arcs that start in ∂T =
f −1 ((Sδ ) \ {0}), Sδ = ∂Dδ , they finish in V and they are transverse to all the
“tubes” f −1 (Sη ) with η ∈]0, δ[.
This vector field ρ defines a C ∞ retraction ξ of T into V , with V as fixed
point set. The restriction of ξ to any fixed Milnor fiber F = f −1 (t0 ) ∩ Bε ,
t0 ∈ Sδ , provides a continuous map π : F → V which is surjective and it is
C ∞ over the regular part of V . As before, we call such map ξ, or also π, a
tube map for V .
We use π to lift the stratified frame v (r) on V to an r-frame ṽ (r) on F.
Given a point x ∈ F, we let γx be the solution of ρ that starts at x. The
end-point of γx is the point π(x) ∈ V . We parameterize this arc γx by the
interval [0, 1], with γx (0) = x and γx (1) = π(x). We assume that this interval
[0, 1] is the arc in Dδ going from to to 0, so that for each t ∈ [0, 1[, the point
γx (t) is in a unique Milnor fiber Ft = f −1 (t) ∩ Bε . The family of tangent
spaces to Ft at the points γx (t) define a 1-parameter family of n-dimensional
subspaces of Cn+1 , that converges to an n-plane Λπ(x) ⊂ Tπ(x) (U ) when t
goes to 1; one has an induced isomorphism Tx F(x)
Λπ(x) .
Since the stratification satisfies Thom’s af -condition, Λπ(x) contains the
space Tπ(x) Vα , tangent to the stratum that contains π(x). Hence, the given
frame v (r) can be lifted to a frame ṽ (r) in Tx F(x). Thus, if the frame v (r)
is defined without singularity on a set A ⊂ V , we obtain an r-frame ṽ (r)
over the inverse image π −1 (A) ⊂ F. The fact that the stratification further
11.3 The Fulton–Johnson Classes 189
satisfies the strict Thom’s wf -condition guarantees that this r-frame ṽ (r) is
continuous.
We summarize this discussion in the following proposition:
In Chap. 3 we stated and proved the Proportionality Theorem for the GSV
index of a vector field. Here we extend that result to r-frames, r ≥ 1. The
method above for lifting a vector field from V to a local Milnor fiber works
for frames. Using the notations introduced in 10.1, we have:
The proof is similar to the one of Theorem 10.2.1 using the liftings defined
in 3.6.1.
A corollary is the following expression of virtual classes:
Theorem 11.3.2. Let us assume that V ⊂ M is a hypersurface, defined by
V = f −1 (0), where f : M → D is a holomorphic function into an open disk
D around 0 in C. For each point a ∈ V , let Fa denote a local Milnor fiber,
and let χ(Fa ) be its Euler–Poincaré characteristic. Then the Fulton–Johnson
r−1 (V ) of V of degree (r − 1) is represented in H2(r−1) (V ) by the cycle
class cFJ
χ(Fai ) I(v (r) , ai ) · σi (11.3.3)
σi ⊂V, dim σi =2(r−1)
r∗ : H∗ (N ) −→ H∗ (V ),
ψ = r∗ ◦ i∗ : H∗ (Vt ) −→ H∗ (V )
the homology specialization map ψ carries the Chern classes of T M |Vt and
N |Vt into the Chern classes of T M |V and N |V , respectively. Thus, as noticed
in [131], one has:
cFJ
∗ (V ) = ψ c∗ (Vt ). (11.3.4)
Let v be, as before, a lifting to Vt via the degenerating map π, of a
(r)
frame v (r) on the 2p-skeleton of V with isolated singularities. The Chern class
satisfying
cp (Vt ) is represented by the obstruction cocycle γ
γ , βi = Obs (
v (r) , Wr (T F)Vt , βi ).
In order to define the Milnor classes in a later section, and to calculate them
in some cases, we develop the localization theory of virtual classes in the
framework of Chern–Weil theory adapted to stratifications and triangula-
tions. Here we use the notation introduced in the previous sections.
We now suppose that V is a compact local complete intersection of
dimension n in a complex manifold M of dimension m, defined as the zero
set of a holomorphic section s of a holomorphic vector bundle N of rank
k = m − n over M . The restriction N |V0 coincides with the normal bundle
NV0 of V0 = V \ Sing(V ) in M and we have an exact sequence of vector
bundles,
π
0 −→ T V0 −→ T M |V0 −→ NV0 −→ 0. (11.4.1)
The virtual class of an r-frame is defined by localizing cp (τV ) by the frame,
where τV denotes the virtual tangent bundle (Sect. 5.1). To be more precise,
let us consider a subset S and suitable neighborhoods as in the previous
sections. Let v (r) be an r-frame on (U \S)∩D(2q) . Let ∇ and ∇ be connections
for T M and N , respectively, on U and set ∇• = (∇, ∇ ). Also, let ∇0 and
∇0 be connections for T M and N , respectively, on a neighborhood W of
(U \ S) ∩ D(2q) in U such that the pair ∇•0 = (∇0 , ∇0 ) is compatible and ∇0
is v (r) -trivial. If we consider the 2q-cochain given by
192 11 The Virtual Classes
cp (∇• ) + cp (∇• , ∇•0 ), γ ∈ C2q (T , ∂ T ),
(D)
γ → (11.4.2)
γ∩R γ∩∂R
where the sum is taken over the connected components of the singular set
Sing(V ) and cr−1 (V ∗ ; v (r) ) is the Chern class of V ∗ relative to v (r) , so that
ι∗ cr−1 (V ∗ ; v (r) ) = i∗ PH(v (r),S).
S⊂Σ
Chapter 12
Milnor Number and Milnor Classes
For most authors, Milnor classes are globally defined as elements in H∗ (V, Z),
on the other hand in [31], these classes are localized at the singular set of
V from the beginning. We explain this in a moment, first we introduce the
global classes; there is one such class in each degree:
∗ (V ) − c∗ (V ))
μ∗ (V ) = (−1)n−1 (cSM FJ
In fact, FJ-classes and SM-classes coincide with the usual Chern classes in
the regular part of V . Thus Milnor classes ought to be concentrated in the
singular set Sing(V ). The results of [31,129] prove that this is indeed the case.
Since the results of [149, 155] prove that in the case of isolated singularities
this contribution corresponds to the local Milnor number at each singular
point, and this is a local invariant of the singularity (not a global one), we
considered in [31] Milnor classes localized at the connected components of
the singular set of V . For each connected component S of Sing(V ), the r-th
Milnor class μr (V, S) of V at S is a homology class in H2r (S, Z) . There is one
such class for each r = 0, 1, · · · , s, where s is the dimension of the component
S. The inclusion S → V maps the homology of S into that of V , and adding
up the contributions in each dimension of all the connected components of
Sing(V ) we get the corresponding global Milnor classes.
For hypersurfaces, the 0-degree localized Milnor class μ0 (V, S) ∈ H0 (S)
Remark 12.2.1. 1. The classes PH(v (r) , S), Sch(v (r) , S) and Vir(v (r) , S) may
be defined for an r-frame v (r) on the intersection of a neighborhood of ∂T
(in V ) and D(2q) , where T = T ∩ V with T a cellular tube around S.
2. If r = 1, i.e., v (1) = (v), PH(v, S), Sch(v, S) and Vir(v, S) are called and
denoted, respectively, the Poincaré–Hopf index IndPH (v, S), the Schwartz
index IndSch (v, S) and the virtual index IndVir (v, S) of the vector field v
[71,111,148,149]. The corresponding Milnor class μ0 (V, S) is a number which
will be discussed in Sect. 12.4.
196 12 Milnor Number and Milnor Classes
In this section, we give a Lefschetz type formula for the Milnor classes at
a nonsingular connected component S of the singular set of V under the
assumption that V satisfies the Whitney condition along S. For the detailed
proof, we refer to [31].
Let U be a tubular neighborhood of S in M with C ∞ projection ρ : U → S.
We set U = U ∩ V and U0 = U \ S and denote by ρ and ρ0 , respectively,
the restrictions of ρ to U and U0 . From the Whitney condition, we see that
the fibers of ρ are transverse to V and that S is a deformation retract of
U with retraction ρ. We identify ρ∗0 (N |S ) with NU0 , and ρ∗ (N |S ) with N |U .
The bundle T ρ of vectors in T U tangent to the fibers of ρ admits a complex
structure, since it is C ∞ isomorphic with the normal bundle of the complex
submanifold S in V . Let T be a (D)-cellular tube around S in U and R a
(D )-cellular tube in T as in Sect. 10.5.2. We set T = T ∩ V and R = R ∩ V
as before.
Let s denote the complex dimension of S and let v (r−1) be an (r − 1)-frame
on the 2(s−r +1)-skeleton S ∩D(2q) of S. In what follows, we set = s−r +1.
(r) (r−1)
By the Schwartz construction, there exists a radial r-field v0 = (v0 , v0 )
on T ∩ D (2q) (r−1)
such that v0 extends v (r−1)
. The radial vector field v0 is
tangent to U0 and possibly has singularities in the barycenters of 2-cells in
S ∩ D(2q) . We may assume that v0 is tangent to the fibers of ρ near ∂ R.
Let v be a vector field on U0 ∩ D (2q)
which is nonsingular and tangent to
the fibers of ρ in a neighborhood U0 of ∂R so that v (r) = (v0
(r−1)
, v) is an
r-frame on U0 ∩ D . For example, the above v0 has these properties.
(2q)
0 −→ T ρ0 −→ T U0 −→ ρ∗0 T S −→ 0,
(r−1)
on W . Also let ∇S be a v0 -trivial connection for T S on a neighborhood
of S ∩ D(2q) . We take connections ∇1 and ∇2 for T U0 so that (∇ρ1 , ∇1 , ρ∗0 ∇S )
and (∇ρ2 , ∇2 , ρ∗0 ∇S ) are both compatible with the above sequence. Thus ∇1
(r)
is v0 -trivial and ∇2 is v (r) -trivial on W . By Lemma 10.5.1, the homology
(r)
class dS (v0 , v (r) ) is determined by
cp (∇1 , ∇2 ) = ci (∇ρ1 , ∇ρ2 ) · ρ∗0 cj (∇S ). (12.3.1)
i+j=p
, U \ S) −−−−→ ∼
H 2q (U H 2 (S)
ρ̂∗
⏐ ⏐
⏐
AM
⏐
(12.3.2)
=
H2r−2 (S) −−−−→ H2r−2 (S),
where the first row is the inverse of the Thom isomorphism, given by inte-
gration along the fibers of ρ, and the second column is Poincaré duality. The
dual of the first row in (12.3.2) gives an isomorphism
, U ∼
\ S) ←−
H2q (U H2 (S),
which shows that every relative 2q-cycle γ (is homologous to a cycle which)
fibers over a 2-cycle ζ of S. By the projection formula, we get from (12.3.1)
(note that the rank of the bundle T ρ0 is n − s):
cp (∇1 , ∇2 ) = cn−s (∇ρ1 , ∇ρ2 ) · c (∇S ),
γ∩∂R ∂Rx ζ
where x is a point in ζ. Noting that the first factor in the right hand side is
d(v0 , v), we proved the lemma, in view of (12.3.2).
(r)
Since IndSch (v0 , S) = cr−1 (S) and IndSch (v0 , x) = 1, from Lemma 12.3.1,
we have the following:
Now we wish to obtain a formula for the virtual class analogous to the one
in Theorem 12.3.3. First, we consider the exact sequence of vector bundles
on U0 :
0 −→ T ρ0 −→ T ρ|U0 −→ NU0 −→ 0. (12.3.4)
198 12 Milnor Number and Milnor Classes
We compute the Chern classes cj (τρ ) of the virtual bundle τρ = (T ρ−N )|U
on U and will see that there is a canonical lifting cjS (τρ ) in H 2j (U, U \ S),
for j > n − s = rank T ρ0 , of cj (τρ ) ∈ H 2j (U ). For this, we consider the
covering U of U consisting of U itself and a tubular neighborhood U 0 of U0
and represent cj (τρM ), τρM = T ρ − N , as a Čech-de Rham cocycle on U (cf.
[102, 156], here we use the notation in [156, Ch.II]).
Let ∇ρ0 be a connection for T ρ0 . Let ∇N be a connection for N |S and
take a connection ∇ρ̂0 for T ρ|U0 so that (∇ρ0 , ∇ρ̂0 , ρ∗0 ∇N ) is compatible with
(12.3.4). Let ∇ˆ ρ̂ be a connection for T ρ on U . We set ∇ρ• ˆ ρ̂ , ρ∗ ∇N )
= (∇
ρ• ρ ∗ N
and ∇0 = (∇0 , ρ0 ∇ ). Then cj (τρ) is represented by a cocycle in A2j (U) =
0 ) ⊕ A2j (U
A2j (U ) ⊕ A2j−1 (U
0 ), where A∗ ( ) denotes the space of differential
forms on the relevant open set, given by
cj (∇• ) = (cj (∇ρ• ρ•
0 ), c (∇ ), c (∇0 , ∇ )).
j ρ• j ρ•
cj (∇ρ• j ρ
0 ) = c (∇0 ),
which vanishes for j > n−s by the rank reason. Thus, for j > n−s, the cocycle
0 ) = {0} ⊕ A2j (U
cj (∇• ) is in A2j (U, U ) ⊕ A2j−1 (U
0 ). Since the cohomology
∗ ∗
of A (U, U0 ) is canonically isomorphic with H (U, U \ S) [156, Ch.VI, 4], this
cocycle defines a class, denoted cjS (τρ ), in H 2j (U, U \ S), which is mapped to
cj (τρ ) by the canonical homomorphism H 2j (U, U \ S) → H 2j (U ). The class
cjS (τρ ) does not depend on the choices of various connections. It should be also
noted that it does not depend on the frames we discussed earlier. Denoting
by A2i (S) the space of 2i-forms on S, we have the integration along the fibers
of ρ [156, Ch.II, 5] ρ∗ : A2(n−s+i) (U, U 0 ) → A2i (S), which commutes with
the differentials and induces a map on the cohomology level :
where ρ∗ and (∂ρ)∗ denote the integration along the fibers of ρ|R and ρ|∂R .
We note that, in the following formulas, the classes ρ∗ cn−s+i
S (τρ ) for i =
1, . . . , k − 1 are involved and they do not appear if k = 1 ( i.e., V is a
hypersurface). We denote by [ ]i the component of degree 2i of the relevant
cohomology class.
12.3 Differential Geometric Point of View 199
k−1 j
∗ −1 ∗
+ cj−i (N ) · ρ∗ cn−s+i
S (τρ ) · c (N ) · c (S) [S].
j=1 i=1
From Theorems 12.3.3 and 12.3.6, we get the following Lefschetz type
formula for the Milnor class.
k−1 j
+(−1) n
c j−i
(N ) · ρ∗ cn−s+i
S (τρ ) · c∗ (N )−1 · c∗ (S) [S],
j=1 i=1
This definition does not depend on the choice of the vector field v and
is equal to μ0 (V, S) in Definition 12.2.1. If (V, a) is an isolated complete
intersection singularity germ, for a radial vector field v0 , IndSch (v0 , a) = 1
and IndVir (v0 , a) = χ(F), where F denotes the Milnor fiber. Thus the above
Milnor number coincides with the usual one in [79, 116, 121].
We recall that the classical Milnor number of an isolated singular point
[121] has been generalized to the case of nonisolated hypersurface singularities
by A. Parusiński [127] in the following way. Recall that a hypersurface V in
M is always defined by a holomorphic section s of a line bundle N over
M . There is a canonical vector bundle homomorphism π : T M |V → N |V
which extends the one in (11.4.1). Note that Sing(V ) coincides with the set
of points in V where π fails to be surjective. Now let ∇ be a connection for
N of type (1, 0). This means that in the decomposition ∇ = ∇(1,0) + ∇(0,1)
of ∇ into the (1, 0) and (0, 1) components, we have ∇(0,1) = ∂. ¯ Since s is
holomorphic, we have ∇ s = ∇ (1,0)
s, which is a C section t of T ∗ M ⊗ N .
∞
μS (V ) = μ(V, S).
Chapter 13
Characteristic Classes of Coherent
Sheaves on Singular Varieties
Abstract As we have seen along this book, for a singular variety V , there
are several definitions of Chern classes, the Mather class, the Schwartz–
MacPherson class, the Fulton–Johnson class and so forth. They are in the
homology of V and, if V is nonsingular, they all reduce to the Poincaré dual
of the Chern class c∗ (T V ) of the tangent bundle T V of V . On the other hand,
for a coherent sheaf F on V , the (cohomology) Chern character ch∗ (F ) or
the Chern class c∗ (F ) makes sense if either V is nonsingular or F is locally
free. In this chapter, we propose a definition of the homology Chern char-
acter ch∗ (F ) or the Chern class c∗ (F ) for a coherent sheaf F on a possibly
singular variety V . In this direction, the homology Chern character or the
Chern class is defined in [140] (see also [100]) using the Nash type modifi-
cation of V relative to the linear space associated to the coherent sheaf F .
Also, the homology Todd class τ (F ) is introduced in [15] to describe their
Riemann-Roch theorem. Our class is closely related to the latter.
The variety V we consider in this chapter is a local complete intersection
defined by a section of a holomorphic vector bundle over the ambient complex
manifold M . If F is a locally free sheaf on V , then the class ch∗ (F ) coincides
with the image of ch∗ (F ) by the Poincaré homomorphism H∗ (V ) → H ∗ (V ).
This fact follows from the Riemann-Roch theorem for the embedding of V
into M , which we prove at the level of Čech-de Rham cocycles. We also
compute the Chern character and the Chern class of the tangent sheaf of V ,
in the case V has only isolated singularities.
We further define the Chern character form and the Todd form by
i
also makes sense. If we set si (∇) = tr(A) , the homogeneous piece of degree
2i in tr(A), then it is a closed 2i-form on M . Denoting by the rank of E,
we have
si (∇)
c∗ (∇) = 1 + ci (∇) and ch∗ (∇) = + .
i=1
i!
i≥1
(−1)i ch∗ (Λi ∇∗ ) = td−1 (∇) · c (∇), (13.1.2)
i=0
q
q
∗
∗ •
c (∇ ) = ∗
c (∇ ) (i) ε(i)
and •
ch (∇ ) = (−1)i ch∗ (∇(i) ),
i=0 i=0
13.1 Local Chern Classes and Characters in the Čech-de Rham Cohomology 203
where ε(i) = (−1)i . If we denote by ci = ci (∇• ) and si /i! = si (∇• )/i! the
homogeneous pieces of degree 2i in c∗ (∇• ) and ch∗ (∇• ), respectively, they
are again related by (13.1.1). More generally, if ϕ = ϕ(c1 , c2 , . . . ) is a series
in ci (we call such a series a symmetric series), we may define a form ϕ(∇• )
(cf. Sect. 5.2). It is a closed form and its class ϕ(ξ) in the cohomology ring
H ∗ (M ; C) is the characteristic class of ξ with respect to ϕ. Suppose further
that we have two families of connections ∇•ν = (∇ν , . . . , ∇ν ), ν = 0, 1,
(q) (0)
• •
for ξ. Then, we have the “difference form” ϕ(∇0 , ∇1 ) satisfying (5.2.1).
Now we state a lemma which will be used to describe explicitly the differ-
ence between the cocycle for the product of two symmetries series and the
product of cocycles for these series. For the proof we refer to [158, Lemma 1.5]
Note that ϕψ(∇• ) = ϕ(∇• )·ψ(∇• ), for symmetric series ϕ and ψ and a family
of connections ∇• .
Lemma 13.1.1. In the above situation, for two symmetric series ϕ and ψ,
we have
where
τ01 = π∗ (ϕ(π ∗ ∇•0 , ∇
˜ • ) · dψ(π ∗ ∇• , ∇
1
˜ • )).
∗
H (M ; C).
From Lemma 13.1.1, we have the following:
where τ = (0, 0, τ01 ) with τ01 a form on U01 as given in Lemma 13.1.1.
ci (∇•0 ) = 0.
In fact, the above holds for the difference form of a finite number of families
of connections compatible with (13.1.3) on U0 . For a symmetric series ϕ
without constant term, we also have a similar vanishing ϕ(∇•0 ) = 0.
Let ∇•0 be a family of connections compatibleq with (13.1.3) on U0 and ∇•1
i
an arbitrary family of connections for ξ = i=0 (−1) Ei on U1 . Then the
class ci (ξ) is represented by the cocycle
in A2i (U). By Lemma 13.1.2, we have ci (∇•0 ) = 0 and thus the cocycle is
in A2i (U, U0 ) and it defines a class ciS (ξ) in H 2i (M, M \ S; C). It is sent to
ci (ξ) by the canonical homomorphism j ∗ . It is not difficult to see that the
class ciS (ξ) does not depend on the choice of the family of connections ∇•0
compatible with (13.1.3) or on the choice of the family of connections ∇•1 .
If ϕ is a symmetric series without constant term, we may also define the
localized class ϕS (ξ) of ϕ(ξ). In particular, noting that the alternating sum
of the ranks of Ei is zero, if M \ S = ∅, we have the localized Chern character
ch∗S (ξ) in the relative cohomology H ∗ (M, M \ S; C), which is sent to ch∗ (ξ)
by the homomorphism j ∗ . It is the class of the cocycle
in A∗ (U, U0 ).
Let E be another vector bundle over M and ∇ a connection for E on M .
Then its Chern character ch∗ (E) is the class of the cocycle
13.1 Local Chern Classes and Characters in the Čech-de Rham Cohomology 205
0 −→ E ⊗ Eq −→ · · · −→ E ⊗ E0 −→ 0
We have
Remark 13.1.2. The local Chern characters defined as above have all the
necessary properties and should coincide with the ones in [86]. Hence they
are in the cohomology H ∗ (M, M \ S; Q) with Q coefficients. Also, the local
Chern classes above are in the image of H ∗ (M, M \ S; Z) → H ∗ (M, M \ S; C).
See also [15] for local Chern characters.
T : H p (V ) −→ H p+2k (M, M \ V ).
A(σ τ ) = i∗ σ A(τ ),
where A denotes the Alexander isomorphism (see Sects. 1.2 and 10.4)
A : H ∗ (U, U \ V ) −→ H∗ (V ).
i∗ : H p (V ) → H p+2k (M ),
T j∗
H p (V ) −−−−→ H p+2k (M, M \ V ) −−−−→ H p+2k (M )
⏐ ⏐ ⏐
⏐P ⏐ ⏐
V A PM
= i
H2n−p (V ) −−−−→ H2n−p (V ) −−−∗−→ H2n−p (M ).
In this and the subsequent sections, we consider the following two cases:
(i) V is nonsingular,
(ii) V is a local complete intersection defined by a section (see Sect. 5.1).
First, suppose V is nonsingular and let p : NV → V be the normal bundle
of V in M . In this case, P and T are isomorphisms. We may take as U above
a tubular neighborhood so that ρ is C ∞ . Then ρ : U → V is isomorphic
with p : W → V for a neighborhood W of the zero section in NV , which we
identify with V . The bundle ρ∗ NV is also isomorphic with p∗ NV . Thus we
have an isomorphism
H ∗ (M, M \ V )
H ∗ (NV , NV \ V ).
The Thom class ΨV of V corresponds to the Thom class ΨNV of the bundle
NV under this isomorphism and the Thom homomorphism corresponds to
∼
the Thom isomorphism TNV : H p (V ) → H p+2k (NV , NV \ V ). Note that, if
we denote by sΔ the diagonal section of the bundle p∗ NV over NV , its zero
set is V and we have ([156, Ch.III, Theorem 4.4])
ΨNV = ck (p∗ NV , sΔ ).
ch∗ (Λ• ∇∗ ) = (ch∗ (Λ• ∇∗0 ), ch∗ (Λ• ∇∗1 ), ch∗ (Λ• ∇∗0 , Λ• ∇∗1 ))
k
whose class in H ∗ (M ; C) is ch∗ (λN ∗ ), λN ∗ = i i ∗
i=0 (−1) Λ N .
Theorem 13.3.1. The cocycle ch∗ (Λ• ∇∗ ) is in A∗ (U, U0 ) and is given by
0 −→ Λk N ∗ −→ · · · −→ Λ1 N ∗ −→ Λ0 N ∗ −→ 0, (13.3.2)
ch∗ (∇F ∗ F ∗ F
) = (ch (∇ ), ch (∇ ), 0)
ch∗ (∇F ∗ • ∗ ∗ F
) ch (Λ ∇ ) = ch (∇ ) td
−1
(∇ ) ck (∇ ) + D(ch∗ (∇F
) τ)
in A∗ (U, U0 )|U .
0 −→ OM (Λk N ∗ ) −→ · · · −→ OM (Λ0 N ∗ ) −→ i! OV −→ 0.
If we compute the local class ch∗V (i! OV ) using this resolution, we see that it
is represented by ch∗ (Λ• ∇∗ ). We have, by (13.1.6),
ch∗ (ρ∗ ξ ⊗ i! OV ) = ch∗ (ρ∗ ξ) ch∗V (i! OV ), in the case (i)
ch∗V (i! F ) = ∗ ∗
∗ ∗
ch (ρ F ⊗ i! OV ) = ch (ρ F ) ch∗V (i! OV ), in the case (ii).
Recall that either ch∗ (ρ∗ ξ) or ch∗ (ρ∗ F ) is represented by ch∗ (∇F
). Recalling
also that N |U
ρ∗ NV and ck (N, s) = ΨV (the Thom class), by Corollary
13.3.1, we get
By (13.2.1), we get the first formula. The second follows from the first.
0 −→ Fq −→ · · · −→ F0 −→ 0,
we have
q
(−1)i ch∗ (Fi ) = 0.
i=0
ch∗ (F ) = ch∗ (F ) [V ].
ch∗ (OV ) = [V ].
c∗ (F ) = γ ∗ [V ],
212 13 Characteristic Classes of Coherent Sheaves on Singular Varieties
then it is not difficult to check that the definition does not depend on the
choice of σ ∗ .
c∗ (F ) = c∗ (F ) [V ].
In particular,
c∗ (OV ) = [V ].
0 −→ IV /IV2 −→ ΩM ⊗OM OV −→ ΩV −→ 0.
s
ch∗ (ΘV ) = ch∗ (τV ) [V ] + τ (V, pi ) [pi ],
i=1
s
c∗ (ΘV ) = c∗ (τV ) [V ] + (−1)n+1 (n − 1)! τ (V, pi ) [pi ].
i=1
13.5 Characteristic Classes of the Tangent Sheaf 213
Recall that the class c∗ (τV ) [V ] coincides with the Fulton–Johnson class
cFJ
∗ (V ) of V , in this case.
Let (V, p) be an isolated complete intersection singularity. If it admits a
good C∗ -action in the sense of [116, 9.B], then τ (V, p) = μ(V, p), the Milnor
number of V at p ([74, 3. Satz], [116, (9.10) Proposition]). On the other hand,
for a variety as in Theorem 13.5.1, the Schwartz–MacPherson class c∗ (V ) of
V is given by a formula in [155] (cf. Theorem 12.2.1). Hence we have
c∗ (ΘV ) = c∗ (V ).
1. M. Abate, The residual index and the dynamics of holomorphic maps tangent
to the identity, Duke Math. J. 107 (2001) 173–207.
2. M. Abate, F. Bracci and F. Tovena, Index theorems for holomorphic self-maps,
Ann. of Math. 159 (2004) 819–864.
3. M. Abate, F. Bracci and F. Tovena, Index theorems for holomorphic maps and
foliations, math.CV/0601602
4. R. Abraham and J. Robin, Transversal mappings and flows, W.A. Benjamin,
Inc., 1967.
5. N. A’Campo, Structures de pseudovariété sur les espaces analytiques complexes,
in Intersection Homology, A. Borel et al. Progress in Mathematics, Vol. 50,
Birkhaüser, Boston, Inc. 1984, 41–45.
6. M. Aguilar, J. Seade and A. Verjovsky, Indices of vector fields and topological
invariants of real analytic singularities, Crelle’s, J. Reine u. Ange. Math. 504
(1998), 159–176.
7. Aleksandrov, A.G. The index of vector fields and logarithmic differential forms.
Funct. Anal. Appl. 39, No. 4, 245–255 (2005).
8. P. Aluffi, Chern classes for singular hypersurfaces, Trans. Amer. Math. Soc. 351
(1999), 3989–4026.
9. V. I. Arnold, Index of a singular point of a vector field, the Petrovskii-Oleinik
inequality and Mixed Hodge structures, Funct. Anal. Appl. 12 (1978), 1–14.
10. V. I. Arnold, S. M. Gusein-Zade and A. N. Varchenko, Singularities of Differ-
entiable Maps, Volume I, Birkhäuser, 1985 (Translated from Russian).
11. M. Atiyah and F. Hirzebruch, Analytic cycles on complex manifolds, Topology,
1 (1961), 25–45.
12. M. Atiyah and F. Hirzebruch, The Riemann-Roch for analytic embeddings,
Topology 1 (1962), 151–166.
13. P. Baum and R. Bott, On the zeroes of meromorphic vector-fields, Essays on
Topology and Related Topics, Springer-Verlag, New York, Heidelberg, Berlin,
29–47, 1970.
14. P. Baum and R. Bott, Singularities of holomorphic foliations J. Differential
Geom. 7 (1972), 279–342.
15. P. Baum, W. Fulton and R. MacPherson, Riemann-Roch for singular varieties,
Publ. Math. IHES 45 (1975), 101–145.
16. Ch. Bonatti and X. Gómez-Mont, The index of a holomorphic vector field on a
singular variety, Astérisque 222 (1994), 9–35.
17. H.-C. Graf von Bothmer, W. Ebeling and X. Gómez-Mont, An algebraic formula
for the index of a vector field on an isolated complete intersection singularity,
Ann. Inst. Fourier (Grenoble) 58 (2008), no. 5, 1761–1783.
18. R. Bott, The stable homotopy of the classical groups, Proc. Nat. Acad. Sci. 13
(1957), 933–935.
215
216 References
115. A. Lins Neto, Complex codimension one foliations leaving a compact submani-
fold invariant, Dynamical systems and bifurcation theory (Rio de Janeiro 1985),
295–317, Pitman Res. Notes Math. Ser. 160, Longman Sci. Tech. Harlow (1987).
116. E.J.N. Looijenga, Isolated Singular Points on Complete Intersections, LMS Lec-
ture Notes 77, Cambridge Univ. Press 1984.
117. R. MacPherson, Chern classes for singular varieties, Ann. of Math 100 (1974),
423–432.
118. D.B. Massey, Lê Cycles and Hypersurface Singularities, Springer-Verlag, Lecture
Notes in Mathematics 1615, 1995.
119. D.B. Massey, Hypercohomology of Milnor fibers, Topology 35 (1996), no. 4,
969–1003.
120. J. Milnor, Topology from the Differentiable Viewpoint, Univ. Press of Virginia,
Charlottesville, 1965.
121. J. Milnor, Singular points of complex hypersurfaces, Ann. of Math. Studies 61,
Princeton 1968.
122. J. Milnor and P. Orlik, Isolated singularities defined by weighted homogeneous
polynomials, Topology 9 (1970), 385–393.
123. J. Milnor and J. Stasheff, Characteristic Classes, Annals of Mathematics Studies
76, Princeton University Press, Princeton, 1974.
124. M. Morse, Singular points of vector fields under general boundary conditions,
Amer. Journ. Math. 51 (1929), 165–178.
125. T. Ohmoto, T. Suwa and S. Yokura, A remark on the Chern classes of local
complete intersections, Proc. Japan Acad. 73 (1997), 93–95.
126. T. Ohmoto and S. Yokura, Product formulas for the Milnor class, Bull. Polish
Acad. Sci. Math. 48 (2000), no. 4, 387–401.
127. A. Parusiński, A generalization of the Milnor number, Math. Ann. 281 (1988),
247–254.
128. A. Parusiński, Limits of tangent spaces to fibers and the wf condition, Duke
Math. Jounal 72 (1993), 99–108.
129. A. Parusiński and P. Pragacz, Characteristic numbers of degeneracy loci, Con-
temp. Math. 123 (1991), 189–198.
130. A. Parusiński and P. Pragacz, A formula for the Euler characteristic of singular
hypersurfaces, J. Algebraic Geom. 4 (1995), 337–351.
131. A. Parusiński and P. Pragacz, Characteristic classes of hypersurfaces and char-
acteristic cycles, J. Algebraic Geom. 10 (2001), 63–79.
132. V. Poenarú, Sur la theorie des immersions, Topology 1 (1962), 81–100.
133. Ch. Pugh, Generalized Poincaré index formulas, Topology 7 (1968), 217–226.
134. C. Sabbah, Quelques remarques sur la géométrie des espaces conormaux,
Astérisque 130 (1985), 161–192.
135. C. Sabbah, Espaces conormaux bivariants, Thèse, Université de Paris VII, 1986.
136. K. Saito, Theory of logarithmic differential forms and logarithmic vector fields,
J. Fac. Sci. Univ. Tokyo Sect. IA 27 (1980), 265–291.
137. J. Schürmann, A short proof of a formula of Brasselet, Lê and Seade, math.
AG/ 0201316.
138. J. Schürmann, Topology of singular spaces and constructible sheaves, Mathe-
matical Monographs-new series, Birkhäuser 2003.
139. M.-H. Schwartz, Classes caractéristiques définies par une stratification d’une
variété analytique complexe, C.R. Acad. Sci. Paris 260 (1965), 3262–3264,
3535–3537.
140. M.-H. Schwartz, Classes et caractères de Chern–Mather des espaces linéaires,
C.R. Acad. Sci. Paris 295 1982, 399–402.
141. M.-H. Schwartz, Champs radiaux sur une stratification analytique complexe,
Travaux en cours 39, Hermann, Paris, 1991.
142. M.-H. Schwartz, Classes de Chern des ensembles analytiques, Actualités
Mathématiques, Hermann, Paris, 2000.
References 221
143. J. Seade, Singular points of complex surfaces and homotopy, Topology 21 (1982),
1–8.
144. J. Seade, The index of a vector field on a complex surface with singularities,
in “The Lefschetz Centennial Conf.”, ed. A. Verjovsky, Contemp. Math. 58,
Part III, Amer. Math. Soc. (1987), 225–232.
145. J. Seade, The index of a vector field under blow-ups, Bol. Soc. Mat. Mexicana
37 (1992), 449–462.
146. J. Seade, Remarks on contact structures and vector fields on isolated complete
intersection singularities. Accepted for publication in the Proceedings of the
IX Workshop on Real and Complex Singularities, to appear in Contemporary
Mathematics.
147. J. Seade and T. Suwa, A residue formula for the index of a holomorphic flow,
Math. Annalen 304 (1996), 621–634.
148. J. Seade and T. Suwa, Residues and topological invariants of singular holomor-
phic foliations, Internat. J. Math. 8 (1997), 825–847.
149. J. Seade and T. Suwa, An adjunction formula for local complete intersections,
Internat. J. Math. 9 (1998), 759–768.
150. J. Seade, M. Tibăr and A. Verjovsky, Global Euler obstruction and polar invari-
ants, Math. Ann. 333 (2005), no. 2, 393–403.
151. J. Seade, M. Tibăr and A. Verjovsky, Milnor numbers and Euler obstruction,
Bull. Braz. Math. Soc., New Series, 36 (2005) 275–283.
152. C. T. Simpson, Higgs bundles and local systems. Inst. Hautes Études Sci. Publ.
Math. No. 75 (1992), 5–95.
153. N. Steenrod, The Topology of Fiber Bundles, Princeton Univ. Press, 1951.
154. T. Suwa, Indices of holomorphic vector fields relative to invariant curves on
surfaces, Proc. of the Amer. Math. Soc. 123 (1995), 2989–2997.
155. T. Suwa, Classes de Chern des intersections complètes locales, C.R. Acad. Sci.
Paris 324 (1996), 67–70.
156. T. Suwa, Indices of vector fields and residues of singular holomorphic foliations,
Actualités Mathématiques, Hermann, Paris, 1998.
157. T. Suwa, Dual class of a subvariety, Tokyo J. Math. 23 (2000), 51–68.
158. T. Suwa, Characteristic classes of coherent sheaves on singular varieties.
Singularities-Sapporo 1998, 279–297, Adv. Stud. Pure Math., 29, Kinokuniya,
Tokyo, 2000.
159. T. Suwa, Residues of Chern classes, J. Math. Soc. Japan 55 (2003), 269–287.
160. T. Suwa, Residues of Chern classes on singular varieties, Singularités Franco-
Japonaises, Marseille 2002, Séminaires et Congrès 10, Soc. Math. France,
265–285, 2005.
161. T. Suwa, Residue Theoretical Approach to Intersection Theory, Proceedings
of the IX Workshop on Real and Complex Singularities, São Carlos, Brazil,
207–261, Contemp. Math., 459, Amer. Math. Soc., Providence, RI, 2008.
162. B. Teissier, Cycles évanescents, sections planes et conditions de Whitney. Singu-
larités à Cargése. Astérisque, Nos. 7 et 8, Soc. Math. France, Paris, p. 285–362,
1973.
163. B. Teissier. Variétés polaires. II. Multiplicités polaires, sections planes, et con-
ditions de Whitney. In Algebraic geometry (La Rábida, 1981), volume 961 of
Lecture Notes in Math., pages 314–491. Springer, Berlin, 1982.
164. R. Thom, Quelques propriétés globales des variétés différentiables, Comm. Math.
Helv. 28 (1954), 17–86.
165. A.K. Tsikh, Weakly holomorphic functions on complete intersections, and their
holomorphic extension, Math. USSR Sbornik, 61 (1988), 421–436.
166. A. N. Varchenko, On the local residue and the intersection form on the vanishing
cohomology, Math. USSR Izvestiya 26 (1986), 31–52.
167. J.L. Verdier, Stratifications de Whitney et théorème de Bertini-Sard. Invent.
Math., 36:295–312, 1976.
222 References
168. J.L. Verdier, Spécialisation des classes de Chern, Astérisque 82–83, 149–159,
1981.
169. S. Yokura On a Milnor class, Preprint 1997.
170. C.T.C. Wall, Topological invariance of the Milnor number mod 2, Topology 22
(1983), 345–350.
171. H. Whitney, Local properties of Analytic Varieties, Symposium in honor of
M. Morse, Princeton Univ. Press, edited by S. Cairns, 1965.
172. V.M. Zakalyukin, Algebraic calculation of the index of a singular point of a
vector field, Funct. Anal. Appl. 6, 69–70 (1972).
Index
223
224 Index
Baum–Bott residues, xvi, 97, 98, 110 ICIS an isolated complete intersection
Bott vanishing theorem, 98, 102 singularity, 46
Camacho–Sad index, xvi, 97, 107, 109 MacPherson class, xviii, 182, 183
cellular tube, 4 Mather class, 182, 183
Chern class, 6 Milnor class, 193, 194
of a bundle, 8 Milnor fiber, 63, 81, 135, 159, 187
of a manifold, 7 Milnor number, xv, 48, 71, 81, 155, 159,
of virtual tangent bundle, 186 161, 193, 194
relative –, 8 morsification, 3
Chern–Weil theory, 11, 86
connection, 11
contact structure, 62 Nash bundle, xviii, 130, 138, 150, 151,
curvature, 12 182
Nash transformation, 130
normally radial, 169
difference cocycle, 9 normally radial 1-forms, 147
dual cell, 4
dual decomposition, 4
obstruction theory, 6
Euler class, 10
Euler obstruction, 53, 130 Poincaré–Hopf class, 9
and hyperplane sections, 133 Poincaré–Hopf index, 1, 4
and the Euler defect, 137, 139 for 1-forms, 146
of a function, 136, 137 of frames, 7, 168
via Morse theory, 140 Poincaré–Hopf Theorem, 2
Euler–Poincaré characteristic, 2 for GSV index, 48
for manifolds with boundary, 2
for Schwartz index, 39
FJ class, 186, 194 Proportionality theorem
frame, 7 for Euler obstruction, 131
Fulton–Johnson class, xix, 185, 186, 189 for 1-forms, 149, 151, 157
for frames, 171
for GSV index, 58, 60
Gauss-Bonnet formula, 94
geometric complete intersection, 74
Grothendieck residues, 104 radial 1-forms, 144, 145
GSV index, 47, 53, 128, 135, 186 radial extension, 34, 168, 175
and the curvatura integra, 80 radial extension of 1-forms, 147, 148
and the Milnor number (real case), radial frame, 170
81 radial index
for 1-forms, 153, 154, 156 of 1-forms, 151–153
on curves with several branches, 49 of vector fields, 32
on germs with nonisolated singulari- radial vector field, 33, 37, 47, 57, 59, 60,
ties, 58 82, 119, 130, 135, 138
on real analytic germs, 73, 74
Vol. 1805: F. Cao, Geometric Curve Evolution and Image Vol. 1825: J. H. Bramble, A. Cohen, W. Dahmen, Mul-
Processing (2003) tiscale Problems and Methods in Numerical Simulations,
Vol. 1806: H. Broer, I. Hoveijn. G. Lunther, G. Vegter, Martina Franca, Italy 2001. Editor: C. Canuto (2003)
Bifurcations in Hamiltonian Systems. Computing Singu- Vol. 1826: K. Dohmen, Improved Bonferroni Inequal-
larities by Gröbner Bases (2003) ities via Abstract Tubes. Inequalities and Identities of
Vol. 1807: V. D. Milman, G. Schechtman (Eds.), Geomet- Inclusion-Exclusion Type. VIII, 113 p, 2003.
ric Aspects of Functional Analysis. Israel Seminar 2000- Vol. 1827: K. M. Pilgrim, Combinations of Complex
2002 (2003) Dynamical Systems. IX, 118 p, 2003.
Vol. 1808: W. Schindler, Measures with Symmetry Prop- Vol. 1828: D. J. Green, Gröbner Bases and the Computa-
erties (2003) tion of Group Cohomology. XII, 138 p, 2003.
Vol. 1809: O. Steinbach, Stability Estimates for Hybrid Vol. 1829: E. Altman, B. Gaujal, A. Hordijk, Discrete-
Coupled Domain Decomposition Methods (2003) Event Control of Stochastic Networks: Multimodularity
Vol. 1810: J. Wengenroth, Derived Functors in Functional and Regularity. XIV, 313 p, 2003.
Analysis (2003) Vol. 1830: M. I. Gil’, Operator Functions and Localiza-
Vol. 1811: J. Stevens, Deformations of Singularities tion of Spectra. XIV, 256 p, 2003.
(2003) Vol. 1831: A. Connes, J. Cuntz, E. Guentner, N. Hig-
Vol. 1812: L. Ambrosio, K. Deckelnick, G. Dziuk, son, J. E. Kaminker, Noncommutative Geometry, Mar-
M. Mimura, V. A. Solonnikov, H. M. Soner, Mathemat- tina Franca, Italy 2002. Editors: S. Doplicher, L. Longo
ical Aspects of Evolving Interfaces. Madeira, Funchal, (2004)
Portugal 2000. Editors: P. Colli, J. F. Rodrigues (2003) Vol. 1832: J. Azéma, M. Émery, M. Ledoux, M. Yor
Vol. 1813: L. Ambrosio, L. A. Caffarelli, Y. Brenier, (Eds.), Séminaire de Probabilités XXXVII (2003)
G. Buttazzo, C. Villani, Optimal Transportation and its Vol. 1833: D.-Q. Jiang, M. Qian, M.-P. Qian, Mathemati-
Applications. Martina Franca, Italy 2001. Editors: L. A. cal Theory of Nonequilibrium Steady States. On the Fron-
Caffarelli, S. Salsa (2003) tier of Probability and Dynamical Systems. IX, 280 p,
Vol. 1814: P. Bank, F. Baudoin, H. Föllmer, L.C.G. 2004.
Rogers, M. Soner, N. Touzi, Paris-Princeton Lectures on Vol. 1834: Yo. Yomdin, G. Comte, Tame Geometry with
Mathematical Finance 2002 (2003) Application in Smooth Analysis. VIII, 186 p, 2004.
Vol. 1815: A. M. Vershik (Ed.), Asymptotic Combi-
Vol. 1835: O.T. Izhboldin, B. Kahn, N.A. Karpenko,
natorics with Applications to Mathematical Physics.
A. Vishik, Geometric Methods in the Algebraic Theory
St. Petersburg, Russia 2001 (2003)
of Quadratic Forms. Summer School, Lens, 2000. Editor:
Vol. 1816: S. Albeverio, W. Schachermayer, M. Tala- J.-P. Tignol (2004)
grand, Lectures on Probability Theory and Statistics.
Vol. 1836: C. Nǎstǎsescu, F. Van Oystaeyen, Methods of
Ecole d’Eté de Probabilités de Saint-Flour XXX-2000.
Graded Rings. XIII, 304 p, 2004.
Editor: P. Bernard (2003)
Vol. 1817: E. Koelink, W. Van Assche (Eds.), Orthogonal Vol. 1837: S. Tavaré, O. Zeitouni, Lectures on Probabil-
Polynomials and Special Functions. Leuven 2002 (2003) ity Theory and Statistics. Ecole d’Eté de Probabilités de
Saint-Flour XXXI-2001. Editor: J. Picard (2004)
Vol. 1818: M. Bildhauer, Convex Variational Problems
with Linear, nearly Linear and/or Anisotropic Growth Vol. 1838: A.J. Ganesh, N.W. O’Connell, D.J. Wischik,
Conditions (2003) Big Queues. XII, 254 p, 2004.
Vol. 1819: D. Masser, Yu. V. Nesterenko, H. P. Schlick- Vol. 1839: R. Gohm, Noncommutative Stationary Pro-
ewei, W. M. Schmidt, M. Waldschmidt, Diophantine cesses. VIII, 170 p, 2004.
Approximation. Cetraro, Italy 2000. Editors: F. Amoroso, Vol. 1840: B. Tsirelson, W. Werner, Lectures on Probabil-
U. Zannier (2003) ity Theory and Statistics. Ecole d’Eté de Probabilités de
Vol. 1820: F. Hiai, H. Kosaki, Means of Hilbert Space Saint-Flour XXXII-2002. Editor: J. Picard (2004)
Operators (2003) Vol. 1841: W. Reichel, Uniqueness Theorems for Vari-
Vol. 1821: S. Teufel, Adiabatic Perturbation Theory in ational Problems by the Method of Transformation
Quantum Dynamics (2003) Groups (2004)
Vol. 1822: S.-N. Chow, R. Conti, R. Johnson, J. Mallet- Vol. 1842: T. Johnsen, A. L. Knutsen, K3 Projective Mod-
Paret, R. Nussbaum, Dynamical Systems. Cetraro, Italy els in Scrolls (2004)
2000. Editors: J. W. Macki, P. Zecca (2003) Vol. 1843: B. Jefferies, Spectral Properties of Noncom-
Vol. 1823: A. M. Anile, W. Allegretto, C. Ringhofer, muting Operators (2004)
Mathematical Problems in Semiconductor Physics. Vol. 1844: K.F. Siburg, The Principle of Least Action in
Cetraro, Italy 1998. Editor: A. M. Anile (2003) Geometry and Dynamics (2004)
Vol. 1824: J. A. Navarro González, J. B. Sancho de Salas, Vol. 1845: Min Ho Lee, Mixed Automorphic Forms,
C ∞ – Differentiable Spaces (2003) Torus Bundles, and Jacobi Forms (2004)
Vol. 1846: H. Ammari, H. Kang, Reconstruction of Small Vol. 1870: V.I. Gurariy, W. Lusky, Geometry of Müntz
Inhomogeneities from Boundary Measurements (2004) Spaces and Related Questions. (2005)
Vol. 1847: T.R. Bielecki, T. Björk, M. Jeanblanc, M. Vol. 1871: P. Constantin, G. Gallavotti, A.V. Kazhikhov,
Rutkowski, J.A. Scheinkman, W. Xiong, Paris-Princeton Y. Meyer, S. Ukai, Mathematical Foundation of Turbu-
Lectures on Mathematical Finance 2003 (2004) lent Viscous Flows, Martina Franca, Italy, 2003. Editors:
Vol. 1848: M. Abate, J. E. Fornaess, X. Huang, J. P. M. Cannone, T. Miyakawa (2006)
Rosay, A. Tumanov, Real Methods in Complex and CR Vol. 1872: A. Friedman (Ed.), Tutorials in Mathemati-
Geometry, Martina Franca, Italy 2002. Editors: D. Zait- cal Biosciences III. Cell Cycle, Proliferation, and Cancer
sev, G. Zampieri (2004) (2006)
Vol. 1849: Martin L. Brown, Heegner Modules and Ellip- Vol. 1873: R. Mansuy, M. Yor, Random Times and En-
tic Curves (2004) largements of Filtrations in a Brownian Setting (2006)
Vol. 1850: V. D. Milman, G. Schechtman (Eds.), Geomet- Vol. 1874: M. Yor, M. Émery (Eds.), In Memoriam Paul-
ric Aspects of Functional Analysis. Israel Seminar 2002- André Meyer - Séminaire de Probabilités XXXIX (2006)
2003 (2004) Vol. 1875: J. Pitman, Combinatorial Stochastic Processes.
Vol. 1851: O. Catoni, Statistical Learning Theory and Ecole d’Eté de Probabilités de Saint-Flour XXXII-2002.
Stochastic Optimization (2004) Editor: J. Picard (2006)
Vol. 1852: A.S. Kechris, B.D. Miller, Topics in Orbit Vol. 1876: H. Herrlich, Axiom of Choice (2006)
Equivalence (2004) Vol. 1877: J. Steuding, Value Distributions of L-Functions
Vol. 1853: Ch. Favre, M. Jonsson, The Valuative Tree (2007)
(2004) Vol. 1878: R. Cerf, The Wulff Crystal in Ising and Percol-
Vol. 1854: O. Saeki, Topology of Singular Fibers of Dif- ation Models, Ecole d’Eté de Probabilités de Saint-Flour
ferential Maps (2004) XXXIV-2004. Editor: Jean Picard (2006)
Vol. 1855: G. Da Prato, P.C. Kunstmann, I. Lasiecka, Vol. 1879: G. Slade, The Lace Expansion and its Applica-
A. Lunardi, R. Schnaubelt, L. Weis, Functional Analytic tions, Ecole d’Eté de Probabilités de Saint-Flour XXXIV-
Methods for Evolution Equations. Editors: M. Iannelli, 2004. Editor: Jean Picard (2006)
R. Nagel, S. Piazzera (2004) Vol. 1880: S. Attal, A. Joye, C.-A. Pillet, Open Quantum
Vol. 1856: K. Back, T.R. Bielecki, C. Hipp, S. Peng, Systems I, The Hamiltonian Approach (2006)
W. Schachermayer, Stochastic Methods in Finance, Bres- Vol. 1881: S. Attal, A. Joye, C.-A. Pillet, Open Quantum
sanone/Brixen, Italy, 2003. Editors: M. Fritelli, W. Rung- Systems II, The Markovian Approach (2006)
galdier (2004)
Vol. 1882: S. Attal, A. Joye, C.-A. Pillet, Open Quantum
Vol. 1857: M. Émery, M. Ledoux, M. Yor (Eds.), Sémi- Systems III, Recent Developments (2006)
naire de Probabilités XXXVIII (2005)
Vol. 1883: W. Van Assche, F. Marcellàn (Eds.), Orthogo-
Vol. 1858: A.S. Cherny, H.-J. Engelbert, Singular
nal Polynomials and Special Functions, Computation and
Stochastic Differential Equations (2005)
Application (2006)
Vol. 1859: E. Letellier, Fourier Transforms of Invariant
Vol. 1884: N. Hayashi, E.I. Kaikina, P.I. Naumkin,
Functions on Finite Reductive Lie Algebras (2005)
I.A. Shishmarev, Asymptotics for Dissipative Nonlinear
Vol. 1860: A. Borisyuk, G.B. Ermentrout, A. Friedman,
Equations (2006)
D. Terman, Tutorials in Mathematical Biosciences I.
Vol. 1885: A. Telcs, The Art of Random Walks (2006)
Mathematical Neurosciences (2005)
Vol. 1861: G. Benettin, J. Henrard, S. Kuksin, Hamilto- Vol. 1886: S. Takamura, Splitting Deformations of Dege-
nian Dynamics – Theory and Applications, Cetraro, Italy, nerations of Complex Curves (2006)
1999. Editor: A. Giorgilli (2005) Vol. 1887: K. Habermann, L. Habermann, Introduction to
Vol. 1862: B. Helffer, F. Nier, Hypoelliptic Estimates and Symplectic Dirac Operators (2006)
Spectral Theory for Fokker-Planck Operators and Witten Vol. 1888: J. van der Hoeven, Transseries and Real Dif-
Laplacians (2005) ferential Algebra (2006)
Vol. 1863: H. Führ, Abstract Harmonic Analysis of Con- Vol. 1889: G. Osipenko, Dynamical Systems, Graphs, and
tinuous Wavelet Transforms (2005) Algorithms (2006)
Vol. 1864: K. Efstathiou, Metamorphoses of Hamiltonian Vol. 1890: M. Bunge, J. Funk, Singular Coverings of
Systems with Symmetries (2005) Toposes (2006)
Vol. 1865: D. Applebaum, B.V. R. Bhat, J. Kustermans, Vol. 1891: J.B. Friedlander, D.R. Heath-Brown,
J. M. Lindsay, Quantum Independent Increment Pro- H. Iwaniec, J. Kaczorowski, Analytic Number Theory,
cesses I. From Classical Probability to Quantum Stochas- Cetraro, Italy, 2002. Editors: A. Perelli, C. Viola (2006)
tic Calculus. Editors: M. Schürmann, U. Franz (2005) Vol. 1892: A. Baddeley, I. Bárány, R. Schneider, W. Weil,
Vol. 1866: O.E. Barndorff-Nielsen, U. Franz, R. Gohm, Stochastic Geometry, Martina Franca, Italy, 2004. Editor:
B. Kümmerer, S. Thorbjønsen, Quantum Independent W. Weil (2007)
Increment Processes II. Structure of Quantum Lévy Vol. 1893: H. Hanßmann, Local and Semi-Local Bifur-
Processes, Classical Probability, and Physics. Editors: M. cations in Hamiltonian Dynamical Systems, Results and
Schürmann, U. Franz, (2005) Examples (2007)
Vol. 1867: J. Sneyd (Ed.), Tutorials in Mathematical Bio- Vol. 1894: C.W. Groetsch, Stable Approximate Evalua-
sciences II. Mathematical Modeling of Calcium Dynam- tion of Unbounded Operators (2007)
ics and Signal Transduction. (2005) Vol. 1895: L. Molnár, Selected Preserver Problems on
Vol. 1868: J. Jorgenson, S. Lang, Posn (R) and Eisenstein Algebraic Structures of Linear Operators and on Function
Series. (2005) Spaces (2007)
Vol. 1869: A. Dembo, T. Funaki, Lectures on Probabil- Vol. 1896: P. Massart, Concentration Inequalities and
ity Theory and Statistics. Ecole d’Eté de Probabilités de Model Selection, Ecole d’Été de Probabilités de Saint-
Saint-Flour XXXIII-2003. Editor: J. Picard (2005) Flour XXXIII-2003. Editor: J. Picard (2007)
Vol. 1897: R. Doney, Fluctuation Theory for Lévy Vol. 1922: A. Friedman (Ed.), Tutorials in Mathematical
Processes, Ecole d’Été de Probabilités de Saint-Flour BioSciences IV. Evolution and Ecology (2008)
XXXV-2005. Editor: J. Picard (2007) Vol. 1923: J.P.N. Bishwal, Parameter Estimation in
Vol. 1898: H.R. Beyer, Beyond Partial Differential Equa- Stochastic Differential Equations (2008)
tions, On linear and Quasi-Linear Abstract Hyperbolic Vol. 1924: M. Wilson, Littlewood-Paley Theory and
Evolution Equations (2007) Exponential-Square Integrability (2008)
Vol. 1899: Séminaire de Probabilités XL. Editors: Vol. 1925: M. du Sautoy, L. Woodward, Zeta Functions
C. Donati-Martin, M. Émery, A. Rouault, C. Stricker of Groups and Rings (2008)
(2007) Vol. 1926: L. Barreira, V. Claudia, Stability of Nonauto-
Vol. 1900: E. Bolthausen, A. Bovier (Eds.), Spin Glasses nomous Differential Equations (2008)
(2007) Vol. 1927: L. Ambrosio, L. Caffarelli, M.G. Crandall,
Vol. 1901: O. Wittenberg, Intersections de deux L.C. Evans, N. Fusco, Calculus of Variations and Non-
quadriques et pinceaux de courbes de genre 1, Intersec- Linear Partial Differential Equations. Cetraro, Italy 2005.
tions of Two Quadrics and Pencils of Curves of Genus 1 Editors: B. Dacorogna, P. Marcellini (2008)
(2007) Vol. 1928: J. Jonsson, Simplicial Complexes of Graphs
Vol. 1902: A. Isaev, Lectures on the Automorphism (2008)
Groups of Kobayashi-Hyperbolic Manifolds (2007) Vol. 1929: Y. Mishura, Stochastic Calculus for Fractional
Vol. 1903: G. Kresin, V. Maz’ya, Sharp Real-Part Theo- Brownian Motion and Related Processes (2008)
rems (2007) Vol. 1930: J.M. Urbano, The Method of Intrinsic Scaling.
Vol. 1904: P. Giesl, Construction of Global Lyapunov A Systematic Approach to Regularity for Degenerate and
Functions Using Radial Basis Functions (2007) Singular PDEs (2008)
Vol. 1905: C. Prévôt, M. Röckner, A Concise Course on Vol. 1931: M. Cowling, E. Frenkel, M. Kashiwara,
Stochastic Partial Differential Equations (2007) A. Valette, D.A. Vogan, Jr., N.R. Wallach, Representation
Vol. 1906: T. Schuster, The Method of Approximate Theory and Complex Analysis. Venice, Italy 2004.
Inverse: Theory and Applications (2007) Editors: E.C. Tarabusi, A. D’Agnolo, M. Picardello
Vol. 1907: M. Rasmussen, Attractivity and Bifurcation (2008)
for Nonautonomous Dynamical Systems (2007) Vol. 1932: A.A. Agrachev, A.S. Morse, E.D. Sontag,
Vol. 1908: T.J. Lyons, M. Caruana, T. Lévy, Differential H.J. Sussmann, V.I. Utkin, Nonlinear and Optimal
Equations Driven by Rough Paths, Ecole d’Été de Proba- Control Theory. Cetraro, Italy 2004. Editors: P. Nistri,
bilités de Saint-Flour XXXIV-2004 (2007) G. Stefani (2008)
Vol. 1909: H. Akiyoshi, M. Sakuma, M. Wada, Vol. 1933: M. Petkovic, Point Estimation of Root Finding
Y. Yamashita, Punctured Torus Groups and 2-Bridge Methods (2008)
Knot Groups (I) (2007) Vol. 1934: C. Donati-Martin, M. Émery, A. Rouault,
Vol. 1910: V.D. Milman, G. Schechtman (Eds.), Geo- C. Stricker (Eds.), Séminaire de Probabilités XLI (2008)
metric Aspects of Functional Analysis. Israel Seminar Vol. 1935: A. Unterberger, Alternative Pseudodifferential
2004-2005 (2007) Analysis (2008)
Vol. 1911: A. Bressan, D. Serre, M. Williams, Vol. 1936: P. Magal, S. Ruan (Eds.), Structured Popula-
K. Zumbrun, Hyperbolic Systems of Balance Laws. tion Models in Biology and Epidemiology (2008)
Cetraro, Italy 2003. Editor: P. Marcati (2007) Vol. 1937: G. Capriz, P. Giovine, P.M. Mariano (Eds.),
Vol. 1912: V. Berinde, Iterative Approximation of Fixed Mathematical Models of Granular Matter (2008)
Points (2007) Vol. 1938: D. Auroux, F. Catanese, M. Manetti, P. Seidel,
Vol. 1913: J.E. Marsden, G. Misiołek, J.-P. Ortega, B. Siebert, I. Smith, G. Tian, Symplectic 4-Manifolds
M. Perlmutter, T.S. Ratiu, Hamiltonian Reduction by and Algebraic Surfaces. Cetraro, Italy 2003. Editors:
Stages (2007) F. Catanese, G. Tian (2008)
Vol. 1914: G. Kutyniok, Affine Density in Wavelet Vol. 1939: D. Boffi, F. Brezzi, L. Demkowicz, R.G.
Analysis (2007) Durán, R.S. Falk, M. Fortin, Mixed Finite Elements,
Vol. 1915: T. Bıyıkoǧlu, J. Leydold, P.F. Stadler, Compatibility Conditions, and Applications. Cetraro,
Laplacian Eigenvectors of Graphs. Perron-Frobenius and Italy 2006. Editors: D. Boffi, L. Gastaldi (2008)
Faber-Krahn Type Theorems (2007) Vol. 1940: J. Banasiak, V. Capasso, M.A.J. Chap-
Vol. 1916: C. Villani, F. Rezakhanlou, Entropy Methods lain, M. Lachowicz, J. Miȩkisz, Multiscale Problems in
for the Boltzmann Equation. Editors: F. Golse, S. Olla the Life Sciences. From Microscopic to Macroscopic.
(2008) Bȩdlewo, Poland 2006. Editors: V. Capasso, M. Lachow-
Vol. 1917: I. Veselić, Existence and Regularity Prop- icz (2008)
erties of the Integrated Density of States of Random Vol. 1941: S.M.J. Haran, Arithmetical Investigations.
Schrödinger (2008) Representation Theory, Orthogonal Polynomials, and
Vol. 1918: B. Roberts, R. Schmidt, Local Newforms for Quantum Interpolations (2008)
GSp(4) (2007) Vol. 1942: S. Albeverio, F. Flandoli, Y.G. Sinai, SPDE in
Vol. 1919: R.A. Carmona, I. Ekeland, A. Kohatsu- Hydrodynamic. Recent Progress and Prospects. Cetraro,
Higa, J.-M. Lasry, P.-L. Lions, H. Pham, E. Taflin, Italy 2005. Editors: G. Da Prato, M. Röckner (2008)
Paris-Princeton Lectures on Mathematical Finance 2004. Vol. 1943: L.L. Bonilla (Ed.), Inverse Problems and Imag-
Editors: R.A. Carmona, E. Çinlar, I. Ekeland, E. Jouini, ing. Martina Franca, Italy 2002 (2008)
J.A. Scheinkman, N. Touzi (2007) Vol. 1944: A. Di Bartolo, G. Falcone, P. Plaumann,
Vol. 1920: S.N. Evans, Probability and Real Trees. Ecole K. Strambach, Algebraic Groups and Lie Groups with
d’Été de Probabilités de Saint-Flour XXXV-2005 (2008) Few Factors (2008)
Vol. 1921: J.P. Tian, Evolution Algebras and their Appli- Vol. 1945: F. Brauer, P. van den Driessche, J. Wu (Eds.),
cations (2008) Mathematical Epidemiology (2008)
Vol. 1946: G. Allaire, A. Arnold, P. Degond, T.Y. Hou, Vol. 1972: T. Mochizuki, Donaldson Type Invariants for
Quantum Transport. Modelling, Analysis and Asymp- Algebraic Surfaces (2009)
totics. Cetraro, Italy 2006. Editors: N.B. Abdallah, Vol. 1973: M.A. Berger, L.H. Kauffmann, B. Khesin,
G. Frosali (2008) H.K. Moffatt, R.L. Ricca, De W. Sumners, Lectures on
Vol. 1947: D. Abramovich, M. Mariño, M. Thaddeus, Topological Fluid Mechanics. Cetraro, Italy 2001. Editor:
R. Vakil, Enumerative Invariants in Algebraic Geo- R.L. Ricca (2009)
metry and String Theory. Cetraro, Italy 2005. Editors: Vol. 1974: F. den Hollander, Random Polymers: École
K. Behrend, M. Manetti (2008) d’Été de Probabilités de Saint-Flour XXXVII – 2007
Vol. 1948: F. Cao, J-L. Lisani, J-M. Morel, P. Musé, (2009)
F. Sur, A Theory of Shape Identification (2008) Vol. 1975: J.C. Rohde, Cyclic Coverings, Calabi-Yau
Vol. 1949: H.G. Feichtinger, B. Helffer, M.P. Lamoureux, Manifolds and Complex Multiplication (2009)
N. Lerner, J. Toft, Pseudo-Differential Operators. Quan- Vol. 1976: N. Ginoux, The Dirac Spectrum (2009)
tization and Signals. Cetraro, Italy 2006. Editors: L. Vol. 1977: M.J. Gursky, E. Lanconelli, A. Malchiodi,
Rodino, M.W. Wong (2008) G. Tarantello, X.-J. Wang, P.C. Yang, Geometric Analysis
Vol. 1950: M. Bramson, Stability of Queueing Networks, and PDEs. Cetraro, Italy 2001. Editors: A. Ambrosetti,
Ecole d’Eté de Probabilités de Saint-Flour XXXVI-2006 S.-Y.A. Chang, A. Malchiodi (2009)
(2008) Vol. 1978: M. Qian, J.-S. Xie, S. Zhu, Smooth Ergodic
Vol. 1951: A. Moltó, J. Orihuela, S. Troyanski, Theory for Endomorphisms (2009)
M. Valdivia, A Non Linear Transfer Technique for Vol. 1979: C. Donati-Martin, M. Émery, A. Rouault,
Renorming (2009) C. Stricker (Eds.), Śeminaire de Probablitiés XLII (2009)
Vol. 1952: R. Mikhailov, I.B.S. Passi, Lower Central and Vol. 1980: P. Graczyk, A. Stos (Eds.), Potential Analysis
Dimension Series of Groups (2009) of Stable Processes and its Extensions (2009)
Vol. 1953: K. Arwini, C.T.J. Dodson, Information Geo- Vol. 1981: M. Chlouveraki, Blocks and Families for
metry (2008) Cyclotomic Hecke Algebras (2009)
Vol. 1954: P. Biane, L. Bouten, F. Cipriani, N. Konno, Vol. 1982: N. Privault, Stochastic Analysis in Discrete
N. Privault, Q. Xu, Quantum Potential Theory. Editors: and Continuous Settings. With Normal Martingales
U. Franz, M. Schuermann (2008) (2009)
Vol. 1955: M. Bernot, V. Caselles, J.-M. Morel, Optimal Vol. 1983: H. Ammari (Ed.), Mathematical Modeling in
Transportation Networks (2008) Biomedical Imaging I. Electrical and Ultrasound Tomo-
Vol. 1956: C.H. Chu, Matrix Convolution Operators on graphies, Anomaly Detection, and Brain Imaging (2009)
Groups (2008) Vol. 1984: V. Caselles, P. Monasse, Geometric Descrip-
Vol. 1957: A. Guionnet, On Random Matrices: Macro- tion of Images as Topographic Maps (2010)
scopic Asymptotics, Ecole d’Eté de Probabilités de Saint- Vol. 1985: T. Linß, Layer-Adapted Meshes for Reaction-
Flour XXXVI-2006 (2009) Convection-Diffusion Problems (2010)
Vol. 1958: M.C. Olsson, Compactifying Moduli Spaces Vol. 1986: J.-P. Antoine, C. Trapani, Partial Inner Product
for Abelian Varieties (2008) Spaces. Theory and Applications (2010)
Vol. 1959: Y. Nakkajima, A. Shiho, Weight Filtrations Vol. 1987: J.-P. Brasselet, J. Seade, T. Suwa, Vector Fields
on Log Crystalline Cohomologies of Families of Open on Singular Varieties (2010)
Smooth Varieties (2008)
Vol. 1960: J. Lipman, M. Hashimoto, Foundations of
Grothendieck Duality for Diagrams of Schemes (2009)
Vol. 1961: G. Buttazzo, A. Pratelli, S. Solimini,
E. Stepanov, Optimal Urban Networks via Mass Trans-
portation (2009)
Vol. 1962: R. Dalang, D. Khoshnevisan, C. Mueller,
D. Nualart, Y. Xiao, A Minicourse on Stochastic Partial Recent Reprints and New Editions
Differential Equations (2009)
Vol. 1963: W. Siegert, Local Lyapunov Exponents (2009) Vol. 1702: J. Ma, J. Yong, Forward-Backward Stochas-
Vol. 1964: W. Roth, Operator-valued Measures and Inte- tic Differential Equations and their Applications. 1999 –
grals for Cone-valued Functions and Integrals for Cone- Corr. 3rd printing (2007)
valued Functions (2009) Vol. 830: J.A. Green, Polynomial Representations of
Vol. 1965: C. Chidume, Geometric Properties of Banach GLn , with an Appendix on Schensted Correspondence
Spaces and Nonlinear Iterations (2009) and Littelmann Paths by K. Erdmann, J.A. Green and
Vol. 1966: D. Deng, Y. Han, Harmonic Analysis on M. Schoker 1980 – 2nd corr. and augmented edition
Spaces of Homogeneous Type (2009) (2007)
Vol. 1967: B. Fresse, Modules over Operads and Functors Vol. 1693: S. Simons, From Hahn-Banach to Monotonic-
(2009) ity (Minimax and Monotonicity 1998) – 2nd exp. edition
Vol. 1968: R. Weissauer, Endoscopy for GSP(4) and the (2008)
Cohomology of Siegel Modular Threefolds (2009) Vol. 470: R.E. Bowen, Equilibrium States and the Ergodic
Vol. 1969: B. Roynette, M. Yor, Penalising Brownian Theory of Anosov Diffeomorphisms. With a preface by
Paths (2009) D. Ruelle. Edited by J.-R. Chazottes. 1975 – 2nd rev.
Vol. 1970: M. Biskup, A. Bovier, F. den Hollander, D. edition (2008)
Ioffe, F. Martinelli, K. Netočný, F. Toninelli, Methods of Vol. 523: S.A. Albeverio, R.J. Høegh-Krohn, S. Maz-
Contemporary Mathematical Statistical Physics. Editor: zucchi, Mathematical Theory of Feynman Path Integral.
R. Kotecký (2009) 1976 – 2nd corr. and enlarged edition (2008)
Vol. 1971: L. Saint-Raymond, Hydrodynamic Limits of Vol. 1764: A. Cannas da Silva, Lectures on Symplectic
the Boltzmann Equation (2009) Geometry 2001 – Corr. 2nd printing (2008)
LECTURE NOTES IN MATHEMATICS 123
Edited by J.-M. Morel, F. Takens, B. Teissier, P.K. Maini
1. Lecture Notes aim to report new developments in all areas of mathematics and their
applications - quickly, informally and at a high level. Mathematical texts analysing new
developments in modelling and numerical simulation are welcome.
Monograph manuscripts should be reasonably self-contained and rounded off. Thus
they may, and often will, present not only results of the author but also related work
by other people. They may be based on specialised lecture courses. Furthermore, the
manuscripts should provide sufficient motivation, examples and applications. This clearly
distinguishes Lecture Notes from journal articles or technical reports which normally are
very concise. Articles intended for a journal but too long to be accepted by most journals,
usually do not have this “lecture notes” character. For similar reasons it is unusual for
doctoral theses to be accepted for the Lecture Notes series, though habilitation theses may
be appropriate.
2. Manuscripts should be submitted either online at www.editorialmanager.com/lnm to
Springer’s mathematics editorial in Heidelberg, or to one of the series editors. In general,
manuscripts will be sent out to 2 external referees for evaluation. If a decision cannot yet
be reached on the basis of the first 2 reports, further referees may be contacted: The author
will be informed of this. A final decision to publish can be made only on the basis of the
complete manuscript, however a refereeing process leading to a preliminary decision can
be based on a pre-final or incomplete manuscript. The strict minimum amount of material
that will be considered should include a detailed outline describing the planned contents
of each chapter, a bibliography and several sample chapters.
Authors should be aware that incomplete or insufficiently close to final manuscripts
almost always result in longer refereeing times and nevertheless unclear referees’ recom-
mendations, making further refereeing of a final draft necessary.
Authors should also be aware that parallel submission of their manuscript to another
publisher while under consideration for LNM will in general lead to immediate rejection.
3. Manuscripts should in general be submitted in English. Final manuscripts should contain
at least 100 pages of mathematical text and should always include
– a table of contents;
– an informative introduction, with adequate motivation and perhaps some historical re-
marks: it should be accessible to a reader not intimately familiar with the topic treated;
– a subject index: as a rule this is genuinely helpful for the reader.
For evaluation purposes, manuscripts may be submitted in print or electronic form (print
form is still preferred by most referees), in the latter case preferably as pdf- or zipped
ps-files. Lecture Notes volumes are, as a rule, printed digitally from the authors’ files.
To ensure best results, authors are asked to use the LaTeX2e style files available from
Springer’s web-server at:
ftp://ftp.springer.de/pub/tex/latex/svmonot1/ (for monographs) and
ftp://ftp.springer.de/pub/tex/latex/svmultt1/ (for summer schools/tutorials).
Additional technical instructions, if necessary, are available on request from:
lnm@springer.com.
4. Careful preparation of the manuscripts will help keep production time short besides en-
suring satisfactory appearance of the finished book in print and online. After acceptance
of the manuscript authors will be asked to prepare the final LaTeX source files and also
the corresponding dvi-, pdf- or zipped ps-file. The LaTeX source files are essential for
producing the full-text online version of the book (see
http://www.springerlink.com/openurl.asp?genre=journal&issn=0075-8434 for the exist-
ing online volumes of LNM).
The actual production of a Lecture Notes volume takes approximately 12 weeks.
5. Authors receive a total of 50 free copies of their volume, but no royalties. They are entitled
to a discount of 33.3% on the price of Springer books purchased for their personal use, if
ordering directly from Springer.
6. Commitment to publish is made by letter of intent rather than by signing a formal contract.
Springer-Verlag secures the copyright for each volume. Authors are free to reuse material
contained in their LNM volumes in later publications: a brief written (or e-mail) request
for formal permission is sufficient.
Addresses:
Professor J.-M. Morel, CMLA,
École Normale Supérieure de Cachan,
61 Avenue du Président Wilson, 94235 Cachan Cedex, France
E-mail: Jean-Michel.Morel@cmla.ens-cachan.fr
Professor F. Takens, Mathematisch Instituut,
Rijksuniversiteit Groningen, Postbus 800,
9700 AV Groningen, The Netherlands
E-mail: F.Takens@rug.nl
Professor B. Teissier, Institut Mathématique de Jussieu,
UMR 7586 du CNRS, Équipe “Géométrie et Dynamique”,
175 rue du Chevaleret,
75013 Paris, France
E-mail: teissier@math.jussieu.fr